Home

100299-d ViewFlex-v28 full 02-01

1. Box Max X Y Min at X Max A id X Centoid Y Box Max X Box Max Y 4 1 i X Max at Y Max Box Max Y The blob points are relative to the entire image calibrated or uncalibrated In addition the center of gravity Centroid X Centroid Y can be calculated in binary or grayscale mode 192 Chapter 10 Analyzing blobs Chained pixels Moments Pixels bordering blobs or delimiting holes are referred to as chained pixels Chained pixels always form a closed chain This implies that the starting pixel in the chain is alsothe closing one f your blob has regions which are 1 pixel wide these pixels are chained twice oncein the forward direction and then in the opposite direction Using Inspector you can only obtain the coordinates of pixels bordering blobs not for holes in blobs To obtain the bordering chains of blobs use the BlobGetOutlineP oi nts function You can usethe chained pixel coordinates tocreatea chain code A chain code is a directional code that records an object s boundary as a discrete set of vectors where each vector points to the next pixel in the chain Using blob analysis operations you can also calculate the moments used to find the center of gravity as well as other grayscale or binary moments You can calculate either central or ordinary moments Central moments use coordinates that are relative to the center of gravity of the blob Ordinary moments u
2. 252 Chapter 15 Calibration 0 0 0 000000 ee Calibration sxe scree Rx Rn 254 Types of distortions 254 Calibrating in Inspector 255 Calibrating your imaging setup 256 Non unity aspect ratio 256 Uniform rotation and scaling 256 Perspective and other spatial distortions 257 Relative coordinate system 260 WAVING su Bae bI a dada gS id eg as gd 262 Displayed world units 262 Returned results 0040 262 Part I Inspector Processing your images Chapter 1 Getting started Ihis chapter presents the features of The Matrox Inspector application It also describes what you need to run Matrox Inspector and the installation procedure 22 Chapter I Getting started Acquiring images Image processing capabilities Welcome to Inspector Inspector is a hardware independent application designed to let you work interactively with images for image capture storage and processing applications Inspector is a 32 bit Windows based package giving you the full potential and ease of use of a graphical interface It is capable of running on any system that runs Windows 95 98 or Windows NT 4 0 With I nspector you can load grayscale and color images from disk in a variety of fileformats or grab them from a wide range of supported input devices They can then be saved to a file exported t
3. CE 2 41 E Mean width Std dev width 0 12 Select Transfer totransfer positional and angleresults to a measurement table n addition edge and stripe centers are now shown in yellow 13 Todetermine whether or not thereare bent pins look at the angle values determined for each stripe in the second Info column column 6 of the measurement table From the measurement data generated for theleft most pin rows one Stripe Pos 1 this pin is clearly bent That is its stripe Demos Example 2 Scaling calibration and pattern matching 61 angle in addition toits two edge angles is not dose enough to 90 degrees as compared to the other two stripes Stripe Pos 2 and Stripe Pos 3 Stripe Pos 1 210 00 18 60 wic 80 57 deg Edgel 221 33 198 67 156 80 04 deg lEdge2 221 36 198864 152 8111 deg 30 210 00 15 67 wic 87 53 deg Edge3 221 49 18851 154 87 31 deg 22148 18851 174 87 86 deg Edge 4 198 Stripe Pos 3 58 210 00 15 85 wic 88 43 deg 221 43 198 51 185 87 48 deg 221 50 188 50 208 89 38 deg Multiple Find 4 1 210 00 21000 3 34 85 85 54 de Statistics Select All If you obtain lines that gooff at undesired angles after thestripe finding operation try one or both of the following a Narrow the search box region a Look atthestripearea that you have selected Ensurethat in this area the pin legs have no artifacts or large dark areas Otherwise
4. Pattern Matching Model dialog box From the Dimension tab you will see the cube that was outlined using the ROI icon From the Search tab click All From the Angle tab check Enable Search With Rotate Then enter 180 for Delta Negative and 180 for Delta Positive this enables a full 360 degree search a Click Search This will build a Measurement Table containing all the positions of the object relative to Pos 1 iol xl a MN PattemPos 1171 40 2215 10000 0 00 deg Paene 176227 5 moe TT TW Measurement Table ViewFlex v 2 8 42 User Manual 0201 3 IMAGE 2 8U Searched Images 8 Save the Pattern Matching Model in the Pattern Folder located in the ViewFlex for SCORBASE root folder 9 Open the Results Table and expand the Pattern Models in the folder tree Check that the name you gave is listed in the tree 10 Use this name when you Find Object ViewFlex v 2 6 43 User Manual 0201 Matrox Inspector version 3 0 User Guide Manual no 10478 301 0300 August 13 1999 Matrox is a registered trademark of Matrox Electronic Systems Ltd Microsoft MS DOS Windows and Windows NT are registered trademarks of Microsoft Corporation Intef Pentium and Pentium If are registered trademarks of Intel Corporation Texas Instruments is a trademark of Texas Instruments Incorporated RAMDAC is a trademark of Booktree All other nationally and internationa
5. Get Yalue In the Variable Name field insert the SCORBASE Variable The cell value will be assigned in the variable In the Row field insert the cell row in the Results Table In the Column field insert the column name in the Results Table The Column can be a number or the exact column name 38 User Manual CT Change Table Change value in any cell of the Results Table Change Table EZ m OK m Cancel In the Row field insert the cell row in the Results Table In the Column field insert the column name in the Results Table The Column can be a number or the exact column name In the Value field insert the SCORBASE variable or constant that you want the cell to have Calibration T Synchronizing the Vision System and the Robot To calibrate real world measurements with the vision system s system of pixels it is necessary to perform the following steps l 2 ViewFlex v 2 8 0201 Execute the HOME routine Open Client Camera Check camera to make sure that the picture is adjusted clear and free of distortions Click the Calibration icon Move robot arm into desired position within the camera s field of vision Using a pen or marker mark a position on a piece of paper or on the coordinate grid close to the robot and strong enough to be seen in the camera picture This position should be aligned with the TCP tool center position the exact center of the robot g
6. 0 161 Masking a model lesen 161 Redefining a model s hot spot 162 Setting a model s search constraints 163 Number of matches 163 Search region 00 00 0000 163 The acceptance level 164 Setting the certainty level 164 The positional accuracy 166 The search speed 0 166 Preprocessing the search model 167 Rotation s no eneen a to ge ied beet e Ea 168 Rotated model search 169 Circular overscan model search 169 Searching for a model within a range ofangles 0 170 Quickly finding the orientation ofanimageormodel 171 Speeding up the search 173 Choose the appropriate model 173 Adjust the search speed constraint 173 Effectively choose the search region and search angle 173 The pattern matching algorithm for advanced users lesen 174 Normalized Correlation 174 Hierarchical Search 176 Search Heuristics 177 Sub pixel accuracy 178 Chapter 10 Analyzing blobs 04 181 Blob analysis 0 0 0 00 0000088 182 Performing blob analysis 183 Blob analysis applications 183 Performing the same blob analysis on different images 00 0 185 Selecting
7. Delta Positive and the default center of rotation used is the center of the search box The approximate location of the edge or stripe is first found by searching at every x degrees within this range where x is specified by the tolerance After the approximatelocation is found I nspector fine tunes the search according to a specified accuracy To be effective you must set the degree of accuracy to a value smaller than that of the rotation tolerance Notethat you can specify which interpolationmodetouse when rotating the search box to the required angles see Chapter 4 for a description of the available modes If searching within a range of angles causes the search box to berotated beyond the edges of theimage indicated by a yellow search box the search box should be repositioned or resized 212 Chapter 11 Finding and measuring edges and stripes Search algorithm Inspector projects the two dimensional search box into a one di mensional line that is it takes the box s profile The pixel summation is performed perpendicular to the search direction Each sum represents the intensity of all the pixels in that column Search box pixel summation Tolocate each edge an edge filter is then applied tothe profile The edge filter first finds the edge value of each profile value The edge value is the difference between one profile value and the ne
8. Intel PC Camera Pro Live Video C Extemal video plugged into camera Optimization Settings Optimizes the exposure color and frame rate settings for the following preference Optimize for maximum frame rate C Optimize for low light conditions C Optimize for best image quality Restore Detauts Help OK Cancel Video Source dialog box Video Source Camera Configuration Main ViewFlex v 2 6 0201 Video Input Choose the video source for the applications that use video Your options for video source include the Intel PC Camera Pro or you can attach an external video device like a VCR or camcorder When you select an external video source some options are not available Optimize for maximum frame rate Ensures that the frame rate never drops below 15 frames per second in dim light the video may appear dark Optimize for low light conditions Ensures that the video is properly exposed in low light conditions like when indoors but may appear washed out in bright light conditions The frame rate for this option is between 8 and 15 frames per second This option is not available when the video source is an external video device Optimize for best image quality Ensures that each frame of the video is high quality but it does reduce the frame rate so that the frame rate never exceeds 15 frames per second 16 User Manual Exposure Color Image ViewFlex v 2 6 0201 Choo
9. Performing a line profile to check lighting conditions To convert the image from color to grayscale usethe Image Convert command Alternatively right click on the image to generate the image s context menu and then select Convert Then choose 8 bit U nsigned from the sub menu The best approach to analyze the noise that is present in the image is to analyze a small region that should have a uniform dark pixel intensity You will usethe statistical tools available in Inspector to do so To determine whether or not the lighting condition is uniform in the small region you will create a thick line profilein the horizontal and vertical direction respectively Performing a lineprofileacross what should beauniformarea will determine ifthelighting condition is uniform in that direction A thick line profile will be generated sothat an average of the pixel values along the width of the line segment will be obtained 1 Tosetthethickness of thelineprofile dick onthelineabove the Options section of the Drawing Tools toolbar The Drawing Tool Properties dialog box appears 2 Ensurethe Outline tab is selected Set the line s pixel width in the Width edit field to 4 You can usethe incremental push button arrows or type in this value directly o 3 Now dick on the Line Profile tool or use the Image Statistics Line Profile command Alternatively you can enable the Profile option and then usethe linetool in the Drawing Tools t
10. To find edge and stripe markers with Inspector you must first define your markers The marker contains the image characteristics toseek within a search box inthetarget image There aretwotypes of markers for which you can search a Edge marker A marker consisting of an edge or multiple edges Edges are sharp changes between two or more adjacent pixels m Stripe marker A marker consisting of a stripe two edges or multiplestripes Stripe marker edges need not be parallel Edge marker Stripe marker Using edge detection techni ques Inspector can find a predominant edge or stripe in a given image area based on certain specified characteristics with sub pixel accuracy nspector records the position and or line of the edge or stripe in the measurement table This information is also displayed in the Result Log window if it is enabled Inspector also allows you to define a multiple edge or stripe marker so that you can search for multiple instances of the sameimagecharacteristics within onesearch box F or example a multiple marker could be used to verify if a series of presumed identical pins are actually of equal width or if the spacing between the pins is within established limits Finally a multiple marker is considered to be only one marker that has a specified number of instances of the same characteristics The search box 209 The search box The search box indicates the area of thetarget image in which tosear
11. Alternatively use the Analysis Pattern New Model command The Model dialog box appears with the Dimensions tab selected and the search model displayed in the upper left corner Deren m9 diced age 54 Chapter 2 A few tutorials A Note for some images it is advisable to mask the undesired areas using the colored eyedropper tool to minimize interference from background pixels and increase the accuracy of your search 4 Now select the Search tab from the Model dialog box Since we are expecting a maximum of three pins in the image type 3 in the Only first edit field Dieke Tazacb Advanced Angie Shier ub tele I7 d FT oDnde nd p jamch Sasha fen EU whois eee i ieee PLU X 5 Toensure that search results are drawn with a bounding box graphic object in theimages overlay right click on the Model dialog box The Model Properties dialog box is presented Select the Draw graphic object in image option under the General tab 6 Click theSearch button to search for thethree occurrences of the model in theimage A few points to consider Example 2 Scaling calibration and pattern matching 55 7 A measurement tableis displayed listing the calibrated positions of all models found in theimagein theX1and Y1 columns and the match score 96 obtained for each occurrence in the Info columns Toone X ODD dag mmt nd 31535 00 deg Donot dosethis tablesincethis p
12. Low positional accuracy can help speed up the search but can also produce match scores that are slightly lower than usual Positional accuracy is alsoslightly affected by thesearch speed If a precise match scoreis important for your application use medium or high accuracy The search speed When searching for a model in an image you can increase or decrease the search speed As you increase the speed the robustness of the search operation the likelihood of finding a match is decreased slightly and match scores might be a little less accurate You adjust the search speed using the drop down list box found in the Search tab of the Model dialog box This command has four settings a High As expected the high speed settings allow the search totake all possible shortcuts performing the search as fast as possible The high speed setting is recommended when searching on a good quality image or when using a simple model since accuracy might be slightly affected otherwise Note the search might have a lower tolerance for rotation when using this setting s Medium Thisisthedefault setting and is recommended for medium quality images or more complex models A search with this setting is capable of withstanding up to approximately 5 degrees of rotation for typical models s Low or Very Low Use these settings only if your image quality is particularly bad and if you have encountered problems at higher speeds Preprocessing the sea
13. Print export See Analyzing images with Inspector Resize Histogram profile windows can be resized maximized minimized and dragged Markers Markers in graph window are vertical bars that you can drag to values and obtain the X and Y or Count in a profile or histogram graph respectively Dynamic updates f you want the graph window to update as you modify the graphic object select Dynamic updates from the graph window s context menu If you deselect this option the graph window will be updated only at the end of a graphic object modification 154 Chapter 8 Histograms profiles and measurements Copy You can use this to copy a histogram or profile graph to a new image Note that when graphs are copied to the dipboard they are copied as normal bitmaps H owever any other images copi ed tothe clipboard from Inspector are copied as DIB bitmaps When pasting the copied image to a new image the DIB bitmaps have priority in Inspector and therefore these images will be pasted intothenew image Tocopy your profile or histogram graph to another image when there are DIB bitmaps copies to the clipboard you need to clear the clipboard of any DIB images Taking measurements using graphic objects Using the Drawing Tools available in Inspector it is possible to take quick on the spot measurements of any object in any type of image When the Measure check box is enabled in the Drawing Tools toolbar all graphic objects drawn in
14. The measurementtable The measurement table displays the results of measurement and pattern matching operations It can also display the positional results of blob analysis if thesearetransferred from theAll Results or One Result table obtained after performing a blob analysis operation Onceresults arein the measurement table calculations can be made and further information extracted using the available measurement operations Notethat theinfor mation that appears under theI nfo columns of the measurement table changes accordingly Working with the measurement table Calibration 143 You can save the measurement table using the File Save command copy its contents to the clipboard using the Edit Copy command export results using the Edit Export commands and delete selected rows using the Delete button To select or deselect a row dick anywhere within the row To select or deselect all rows dick on the Select All or Unselect All button respectively To modify a column s width place your cursor over the column s border and drag it to the required width Calibration Using Inspector you can calibrate your imaging setup so that pixel coordinates map to world coordinates Calibrating your imaging setup allows you to obtain analysis results in world units view the current cursor position in world coordinates in status bar of the image and if necessary physically correct an images distortions In general by getting res
15. connect them together 2 Insert the Inspector CD into a CD drive The Setup utility should run automatically If the Setup utility does not run automatically dick on the Start button select Run and then type at the prompt using the appropriate drive letter d setup If Inspector is run without the hardware key a temporary licenseis assigned to your system allowing use of I nspector for 30 days Each time you run Inspector a dialog box appears indicating the number of days until the temporary license expires Oncethis time period has elapsed Inspector will not run unless a hardware key is attached Note that I nspector s 30 day evaluation license can only be installed once Any attempt totamper with theP C s calender before the date of expiry will disable nspector In that event Inspector can only be re used once a hardware key is obtained Once Inspector is installed a group titled Matrox Inspector 3 0 containing the icons for the Inspector application is created Note that one of the icons is for the readme chm file This is an on line help file containing important information about troubleshooting and features specific to imaging boards Be sure to read it The following are also created after installation m Anentry for Inspector in the Windows registry This contains information used by Inspector during start up m Directories containing image sequence C Script Basic script utility and sample autom
16. for example lighting and angular displacement A collection can help you keep track of these models and images F or information on collections see Chapter 3 Collection Files Steps to performing a search 159 Models defined in Inspector can be used in other pattern matching applications that have been built with MIL development tools See the pattern scr demo for an example of a simple pattern matching application Steps to performing a search The basic steps to search for a model in an image are 1 Define your search model You can manually create the model or you can have Inspector automatically select a unique one for you 2 Fine tunethe model This might involve masking regions of the model and or redefining the model s hot spot reference point 3 Set the model s search constraints 4 Preprocess the model If required save the model 5 Search for the model in an image Once you save a model you can load it from disk for future searches In such a case you won t need to perform steps 1 to 4 unless the model requires adjustment Note that in some cases the orientation of a model and or target image can play a factor in the search operation To simplify matters these issues are discussed after the basic Steps are covered under the Rotation section The define and search operations can only be performed on 8 bit images If you try to perform these operations on an imagethat is not 8 bit Inspecto
17. out on the main pane DICOM images in particular have much information that can bedisplayed in theoverlay When playing back a sequence with alot of data in the overlay you might want to check the Optimize grab with overlay feature in the Options Preferences command This option when checked optimizes the displayed sequence so that the overlay does not flash as each frame is played Although this option improves the sequence s playback it has a slight effect on the playback speed 246 Chapter 14 Creating and manipulating sequences Grabbing sequences To grab a sequence you must have your digitizer and camera correctly configured Once you have configured you can grab using the File Grab Sequence command or select the Sequence button from the Digitizer toolbar You must select the type of sequence you wish to grab either a memory type or a disk type sequence as well as the number of frames to grab These types of sequences will be discussed later You can also specify the size of the grabbed frames and the rate at which to grab frames You also have previewing options when grabbing sequences In the Timed Acquisition Setup window there aretwo options Preview and Preview while grab The Preview button opens a paneand allows you tosee what your camera will grab If the Preview while grab checkbox is checked the preview window will open and remain open during the grab Note that while previewing while grabbing is necess
18. range from O to 255 Adjust the pixel intensity Input Range to between 50 and 175 using the slider or type these values directly intothe edit fields A live preview is obtained c mj mm m m 4 bk m jj se wm 4 Now dick the Lock Live Preview button to lock the on screen display in preview mode Alternatively usethe OptionsLock Preview command Thelockedlivepreview allows you to preview the remapped pixel range in the source region during playback without making permanent changes to the pixel range Closethe dynamic histogram graph to allow for a faster playback rate Click on the Repeat button and Play the sequence again Example 3 Working with many images and non standard image file types 67 6 Now adjust the Input Range to between 75 and 160 duringthe playback whilethe preview is maintained This pixel range should provide an optimal contrast level acer espeie ze me Dem CILE ET EJF ALSP u Da ij s Mesa anigeed HN 7 Click on the Stop button after a few iterations 8 Now unlock the Live Preview by clicking on the lock button again The unlocked icon reappears Adjusting the playback By viewing a limited range of frames in the angiogram range in a sequence sequence it is possibleto determine that the point of maximal dynamic flow from the ruptured vessel occurs in frames 6 10 and an instance of maximal displacement for the main vessel occurs in frame 28
19. reposition the search box to avoid these areas Toseemore examples of calibration operations in Inspector run the Calibration and I mage Correction demos For an example of astripefindingand pattern matching operation respectively run the Broken Gear and Track Object demo respectively To run any of these demo 1 UsetheFile Open command or Explorer to open the demo launcher BasicDemo bas that can be found in the V BasicFiles folder Alternatively use the Options Utilities Basic Demo command ChoosetheWaitfor user response option torun thedemo Select the demo and then click the OK button 62 Chapter 2 A few tutorials The Calibration Calibration bas demo is a Inspector script that has been written to physically correct a radially distorted image It first determines the correction needed by calibrating the imaging setup using a grid image and then applying the correction to the destination image Similarly the Image Correction demo CalibGrid bas is an Inspector script that has been written to physically correct an image with a distorted perspective by first calibrating the imaging setup using a grid TheBroken Gear demo BrokenGear bas is an Inspector script that incorporates a multiple stripe find operation to detect the broken teeth of a gear TheTrack Object demo Tracki ngDemo bas is a script that uses a pattern matching operation to track a moving object in a sequence file This demo will also serve as
20. 0 center point and the directions of X and Y are now the same as the robot The movement of the mouse along the axes shows changes in X and Y Click OK Now the Vision System is calibrated and synchronized with the robot and all new images will become calibrated Opening and Saving Calibration Files The Configuration Step 2 dialog box enables file management operations Save Saves the calibration The file name extension is CAL Open Allows you to load backup calibration files When the OK icon is clicked the system is calibrated You can configure immediately in the Step 2 tab and open the calibrated file Example of Creating a Pattern Model ViewFlex v 2 6 0201 l System must be calibrated so that real world measurements correspond with the vision system s set of pixels Open Client Camera 3 Click the Image Processing Tool icon to open Matrox Inspector For this example set a cube down within the camera s field of vision so that its edge is parallel to the edge of the Client Camera window In order for both the robot gripper and the cube to be set to zero degrees the cube must be defined as parallel to the X axis of the camera s world Subsequently the robot arm s rotation will be relative to this point 5 In Client Camera click Snap Click the ROI Region of Interest icon and outline the perimeter of the cube that was set 41 User Manual 7 Click the Pattern Matching icon icbxj
21. 03 43 2 3 33 120 120 140 140 Source image resized 200 in X amp Y using no interpolation Source image a Bilinear Inthis mode Inspector averages the values around a point and copies that average into its associated position Bilinear interpolation produces more accurate results than nearest neighbor interpolation but can take three ti mes as long to perform m Bicubic This modeis like bilinear interpolation except that Inspector uses more pixel values to determine the average Bicubicinterpolation produces the most accurateresults but can take 8 10 times longer to perform than nearest neighbor interpolation Chapter 5 Image transformations This chapter describes basic transformation operations that can be performed on your images 116 Chapter 5 Image transformations Binarizing Band Pass or Two Level Operating on your images Once you haveimproved your image using previously discussed techniques you are ready to start operating on your image to meet your specific needs This chapter discusses image enhancements and transformations The following chapter discusses advanced image analysis Thresholding images Using the Image Processing Threshold command you can threshold your image Thresholding maps all pixels in a specified intensity range to a new constant value Since a thresholded image contains fewer pixel valuesthan theoriginal image some ope
22. 1 Adjust the range of frames in the sequence for playback to Land 30 by right clicking on the dual slider and dragging A small yellow rectangular text box is displayed showing the frame numbers to be included in the sequence Alternatively you can select therangeof frames for viewing do jl by right clicking on the image and then choosing Properties from the sequence s context menu 2 Inthe Properties dialog box select the Sequence I nfo tab In the Playback Range section type 1and 30 in the In and Out edit fields respectively 3 Click on Repeat and Play to play this limited sequence Creating a memory Now you will create a sequence made up of this limited range sequence of frames Since you cannot modify the frames of a sequence that is saved to disk known as a disk sequence you must 68 Chapter 2 A few tutorials create a memory sequence A memory sequencecan be changed because each frameis allocated its own memory Then you can apply the window leveling to this range of frames in the sequence to obtain a better contrast between all blood vessels and the background 1 2 aun RB w Use the File New command Select Sequence from the New Inspector Documentlist box Note donot select SequenceAVI or Sequencel F because these create disk sequences that is each frame added to either of these sequences is saved immediately to disk Therefore the frames of these sequences cannot be modified only rea
23. 44 Chapter 2 A few tutorials Removing random noise from an image using a filter 4 Position the graphs in Inspector s workspace in a tiled fashion sothat they areall easily visibleon screen Ensure that the image is also visible You can change the height and width of the histogram window iR iMai jea Badrilal AL du liae Although the noise in the image leads to variation in pixel intensity the distribution is relatively random therefore the noisein theimagecan beremoved using a smoothing filter You will determine the number of ti mes to apply the filter based on the histogram of the flat area since a narrow histogram curve will indicate that there is a more uniform area 1 Click on the Filters tool button or usethel mage ProcessingFilterscommand Alternatively right click on the image to obtain the image s context menu and select Processing The Processing Operations dialog box is presented 2 Ensurethat theFilters tab is displayed Select Smooth5x5 from the list box Example 1 Analyzing a blister package 45 3 Makesurethe destination D Selector bar at the bottom of the Processing Operations dialog box is set to R BLISPACA J PG not New 3 Band RGB 4 Click on Apply twotimes to perform the smoothing operation twice Note how the graphs changes each time Fromthe change in the histogram of the small region you can seethat the background noiseis reduced and the region has a more uniform value
24. Chapter 11 Finding and measuring edges and stripes The width of a stripe marker is the average distance in pixels between its edges Note if the marker s search box is at an angle the width is measured according tothe orientation of the box as shown below Search box Edge strength A marker s edgestrength istheminimum maximum edgevalue along the width of the edge depending on the polarity of the edge The edge value is represented as a normalized percentage of the maxi mum pixel value possiblefor the specific image buffer Thesign of theedge valuerepresentsthe polarity of the edge For example for a search box with a vertical orientation the equation for an edge value is edge value A adjacent profile values box height abuts depth 1 100 For instance in an 8 bit image buffer the maximum possible pixel valueis 255 Therefore a 50 edge strength in this buffer represents a maximum difference in average adjacent profile values of 128 and arising edge The larger the absolute edge value the greater the edge strength The default setting finds the marker with the largest edge strength Markers advanced characteristics 217 The edge in the diagram below has an edge strength of 100 the maximum edge value possible since the edge is completely vertical and has an edge width of one pixel Background marker pixel pixel value 0 value 255 bod ae Ee ee ae ee eg eee EE Pixel summation Se
25. Fail Sends fail to the Manager Simulation Simulation Mode allows you to simulate test results by determining the Pass Fail according to the percentage of failure that was input to the OpenCIM Manager Debug The Debug Mode allows you to debug every step of the script This is made for testing the script before it will be run online The following are activated only in Debug mode al Open File lets you load the script Run Script icon executes the script that has been chosen Step By Step icon lets you to execute each command line of the script l individually Example of Script for Pass Fail Test ViewFlex v 2 8 0201 The following script is an example for testing Pass Fail Models A Model refers to the pattern for which you are searching and the image for which it is extracted The following program will try to find the x object by using the x model x mod If found Fail is sent QCR Fail If not found then it will try to find the v object If found Pass is sent QCR Pass If neither is found Error is sent QCR Error For more information on the commands refer to Inspector s online help Function QCR As String Dim x As Integer IMAGE1 Insptr ImgGetCur Insptr ImgSetCurrent IMAGE1 R_Def ALL BANDS X mod is a Model Search of object for Fail sign M X MODS Insptr PatLoad C OpenCIM32 MICROUSA WS3 X mod Insptr PatSetCur M X MOD Insptr ImgConvertType TO 8U Insptr
26. MIN 1 is the minimum of all pixels in the neighborhood that correspond to 1 in the kernel MATCH Determine how many elements of the neighborhood match the corresponding value and set the center pixel to this number OPEN Binary erode then binary Grayscale erode then grayscale dilate dilate using the same using the same kernel kernel CLOSE Binary dilate then binary Grayscale dilate then grayscale erode erode using the same using the same kernel kernel HIT OR MISS f a pixel s neighborhood matches the kernel exactly set it to 1 otherwise set it to 0 130 Chapter 6 Advanced image processing Fast Fourier transformations To identify consistent spatial patterns in an image which can be caused for example by systematic noise you can perform afastFouriertransformusingthe ImageFourier Transform commands FFTdefined A fast Fourier transform of an image can be interpreted as the decomposition of the imageintoa set of basic 2 D patterns The composition of these patterns make up the original image The forward Fourier transform is defined as N 1 M 1 F u v Y MID SE exp SE x 0 y 0 where u and v are coordinates in the frequency domain and x and y are coordinates in the spatial domain A forward FFT yields a real R and an imaginary I component of the image in a frequency domain spectrum Theinverse Fourier transform is defined as N 1 M 1 fG y L Y Y F u v exp 203290 e
27. MeasNew x Insptr PatFind If x 1 Then QCR Fail 32 User Manual Else V mod is a Model Search of object for Pass sign M Y MOD Insptr PatLoad C OpenCIM32 MICROUSA WS3 V mod Insptr PatSetCur M Y MOD x Insptr PatFind If x 1 Then QCR Pass Else QCR Error End If End If Insptr PatClose Insptr ImgClose Insptr MeasClose Insptr CloseAll End Function ViewFlex v 2 6 33 User Manual 0201 3 ViewFlex for SCORBASE ViewFlex for SCORBASE Toolbar Results Table ViewFlex v 2 6 0201 The ViewFlex Toolbar contains icons that are shortcuts for selecting certain commands and functions i ViewFlex for Scorbase PSI x A RBS T S ViewFlex Toolbar e Image Processing Tool see details in Chapter 2 e Camera Client Camera for ER 4pc only see details in Chapter 2 e Results Table e Calibration e About e Exit The Results Table window is divided into two sections one section is a tree representing the ViewFlex folders and the second section is a table with the coordinates of the object in the robot world including blob features 34 User Manual File Results Table File Pattern Option View 5 ViewFlex Folders Postion UE ES Dou ur Label Aen Feret Hin Ange Num a Script Calibration Pattern Models Blob Setting B B 5 245018 148 146 D 90 143 E Blobseti E BlobTest Results Table Every time
28. Polarity The polarity of an edge or a stripe s edges is rising or falling A rising edge denotes a risein grayscale values and a positive polarity A falling edge denotes a decrease in grayscale values and a negative polarity When setting the polarity of a marker itisimportanttokeepin mindthedirection of thesearch which is indicated by the search box s search direction line Theedges of a stripemarker can haveopposite polarities or can havethe same polarity By default the polarity is set to Auto In this case Inspector searches for a strong edge or stripe without considering its polarity Position The marker s position is defined as the X and Y coordinates of the marker s center the center of the portion of the marker located within the search box These coordinates are relative to the top left pixel of the image and are returned to the 214 Chapter 11 Finding and measuring edges and stripes measurement tableas the reference position of the marker The reference point coordinates of the marker respect calibration settings on output Theposition of a stripe marker is considered the center between the two edges of the stripe Reference position When several edges have similar characteristics within the same search box you can usethe location lineto specify the approximate position at which tofind the required marker s reference position Contrast You can indicate the typical difference in grayscale values betwe
29. a new file is saved in one of the ViewFlex for SCORBASE folders the tree is updated e Selecting one of the Pattern Models executes Find Object with the model on the current image e Selecting one of the Blob Settings executes Find Blobs with the Blobset on the current image e Selecting one of the Scripts executes the first function subroutine in the Script Save Results to file so that they can be opened with Excel Load results from file into the Results Table Pattern Option View ViewFlex v 2 8 0201 Foreground Defines whether black pixels or white pixels are to act as the foreground or blob pixels With Blob Features Adds the blob feature to the robot coordinate in the table This applies only to Find Object Always on top Checking this box prevents this window from being hidden by other windows 35 User Manual Vision Commands in SCORBASE All SCORBASE programming commands are available from the Command List and are easily compiled into a robot program The list shows the two letter hot key combinations that allow you to enter commands from the keyboard Many commands open dialog boxes for completing the command line parameters Command List Iof x RS ReturnSubroutine Command CS CallSubroutine T Eo LA Label J AXIS CONTROL PS PrintToScreenL og Q PROGRAM FLOW INPUTS amp OUTPUTS c INPUTS amp OUTPUTS I Ifinputit OnJump VISION COMMANDS
30. a new image Notethat you can usean existing image as the destination by using a script function When processing images you often need to perform such color conversions because m All operations under the Analysis menu can only be performed on 8 or 16 bit unsigned grayscale images If your source is not 8 or 16 bit unsigned an 8 bit unsigned image will be created automatically when using a command under the Analysis menu A specific color space might be better suited to a particular application and converting tothat color space minimizes the amount of processing For example you might want to determinehow many basiccolors arerepresented in an image 92 Chapter 3 Dealing with images that contains several colored objects some of which arein a shadow If theimageis an RGB color image all three components must be examined before classifying a color If however thesameimageis represented in HSL only thehue component need be analyzed to determine a color Portions of an object lying inside and outside of a shadow havethe same hue although their luminance and possibly their saturati on values are different Note that if you perform an arithmetic copy operation using different types for the source and the destination the data is simply copied without being scaled or converted to the destination When converting to a smaller buffer depth Inspector attempts to scale the data down as best as possible For a summary of ho
31. also allows you to examine a range of interest in more detail For example if your image produces a bi modal histogram you might want to stretch both ranges to examine them in detail To perform a window leveling operation use the Image Processing Mapping Brightness Contrast Window leveling Histogram equalization command Adjusting the intensity distribution 107 Adjusting the brightness and contrast linearly Using the Image Processing Mapping Brightness Contrast Wi ndow leveling Histogram equalization command you can adjust the brightness and contrast in your image The brightness and contrast slide and stretch the intensity distribution respectively of an image relative to the center of the intensity range The following diagram shows how changes to the brightness and contrast affect the image intensity mapping 2554 2554 T 192 A lt c ntrast cr brightness contrast Va J ut v 0 0 gt 64 128 192 255 64 128 192 255 Original pixel values original pixel values m o Ds 128 brightness new pixel values new pixel values Note that a 100 increase in contrast moves all values above the middle intensity to the maximum intensity and all those below the middleintensity to zero A 100 decrease reduces an image to a single value the middle value of the destination image s full dynamic range 108 Chapter 4 Improving image quality Histogram equalization mapping Histog
32. and hencethe accuracy is half of what can be attained at level 0 Selecting low accuracy might cause the search to stop at level 2 sothe accuracy is reduced by an additional factor of two to about 0 2 pixel 180 Chapter 9 Pattern matching Chapter 10 Analyzing blobs This chapter describes how to perform a blob analysis operation 182 Chapter 10 Analyzing blobs Segmentation Source and mask images Blob analysis Blob analysis is a branch of image analysis that allows you to identify aggregated regions of pixels known as blobs within a grayscale image Once these regions are identified you can calculate selected features of these regions automatically discard regions that are not of interest and classify the remaining regions according to the values of the features Inspector must be able to identify these regions as blobs This is done by segmenting the image so that the regions to be recognized as blobs are in the same logical pixel state foreground state Pixels not part of a blob make up the image s background The segmented image is known as the segmentation mask also Known as the blob identifier image Inspector uses the segmentation mask for both the identification of blobs and the computation of their binary features F or grayscalefeatures the segmentation mask is only used for the identification of blobs while the source region s pixel values are used to calculate the feature By default whe
33. axes This command is entered from keyboard but does not require Enter for execution 24 User Manual Go To Position ex Go to position Go to position Movement Table Position Z Axis mm Move Current fi jeo Fixed angle Dk Cancel Go to Position dialog box Movement Move MoveL Move Linear Table Can be the Current table or any other named measurement table Position The index of a position in the measurement table Z axis mm The Cartesian coordinates in millimeters Fixed Angle Enabling this option causes only the X and Y axes to be This option can be used with circular objects when the gripper rotation is not important Example of Using Go To Position ViewFlex v 2 6 0201 1 System must be calibrated so that real world measurements correspond with the Vision System s world of pixels Open Camera 1 Click the Image Processing Tool icon In this example put a cube down within the camera s field of vision so that its edge is parallel to the edge of Camera 1 dialog box s border In order for both the robot gripper and the cube to be set to zero degrees the cube must be defined as parallel to the X axis of the camera s world Subsequently the robot arm s rotation will now be relative to this point In Camera 1 click Snap Click the ROI Region of Interest icon and outline the perimeter of the cube that was set Click the Pattern Matching
34. blob analysis applications requireonly a few simplesteps whereas others require most of the blob analysis functionality included in Inspector The image you start with is easy to segment if the grayscale values of the objects are distinct from those of the background Toensurethis usea background with very different grayscale values from those of the objects when acquiring the image When the grayscale values of the objects are not distinct from the background you can use the processing operations to segment the image and then the segmented image can be used in the blob analysis by importing it as a segmentation mask Blob analysis applications The following are some typical blob analysis applications Presence detection The purpose of this type of blob analysis is to count all blobs in an image usually one type of blob and to calculate some blob features which can then be used for example to determine each blob s orientation If the calculated features are position type features such as the centroid of a blob their values can then betransferred to a measurement table for further analysis Advanced presence detection The purpose of this type of blob analysis is to count each type of blob in an image two or more To do so each blob first has to be identified by type using classification 184 Chapter 10 Analyzing blobs Inspection of mechanical parts The purpose of this type of blob analysis is to identify f
35. blob features 0 186 Area and perimeter 186 Dimensions a iaa 187 Determining the shape 189 Segmenting an image ses 193 From a source image s 193 Importing a segmentation mask 193 Specifying your segmentation preferences 194 Counting and calculating selected blobfeatures llle 196 Saving and loading blob settings 197 Filtering blobs based on features 198 Definingafilter lessen 198 Classifying blobs 200 Bins method ius s has Eo ay ER 201 Filters method llle 201 Training set method 201 Statistics and tolerances 203 Chapter 11 Finding and measuring edges and stripes Finding and measuring edges and stripes Steps to finding and obtaining measurements of markers Markers edges and stripes The search box llle Searching within a range of angles Search algorithm lees Markers fundamental characteristics Polarity ees ss Bek Ree ate ee uns POSTION rs be BAe Ra yon hee Bes Edge strength leeren Weight factors 0000 Inside edges e ees Multiple marker characteristics Measurement results less Chapter 12 Defining ROIs and annotating images 223 Using ROIs and graphic an
36. classification options see the on line procedure Setting classification options Classifying blobs 201 You can view the statistics for all blobs or for only those that belong to a specific class Refer to the Statistics and tolerances section Bins method The bins method is used to sort blobs into several sub classes of a single feature main dass In general the blobs in the image cover a wide range of values in the selected feature To define a bins dass you must select a blob feature which defines the main class and then specify the range of values of this feature for each required bin sub class For more information see the on line procedure Bins method Filters method Thefilter method is used when a known combination of features can be used as criteria to determine which blobs should be included in the class To define a filter class you must specify the value or range of values of the selected blob feature s to use as criteria for filtering blobs The blobs that meet the specified criteria are then included in the class Define class filters the same way you would define blob filters refer tothe Filtering blobs based on features section The only difference in the class filters method is that the blobs passing the criteria are included in the class For more information see the on line procedure Filter method Training set method Thetraining set method is used when you can determine by looking at the blo
37. colors it considers similar This process is time consuming and must be done whenever a particular imageis Selected To save processing time Inspector maintains a Common palette which when used tells the palette manager to usethe same palette the Common palette for all images The palette manager tries to match the images colors as closely as possible with those availablein the palette The Common palette contains several shades of red green blue yellow for color images and several shades of gray for single band grayscale and therefore provides adequatedisplay colors for most images and should suit most of your application needs This option provides the fastest image repainting Using palettes with an 8 bit display resolution 101 a Own palette Sometimes the Common palette is unsuitable for example when the images ideal palette has few color in common with the common palette or when grabbing directly into the VGA for example with M atrox Meteor Il You can use the Own palette option in such a case which tells the palette manager to use the currently selected image s palette for display purposes and displays this image with the best quality possible The palette manager then triestoprovidethedosest color match possible for the background images images in windows other than the active window if background repainting has not been disabled However with the Own palette option the background images may not display properly
38. corresponding frequency images contain a single point of brightness spike with its associated negative frequency mirrors Note that the spikes in the frequency domain appear in the direction of the pattern The distance between the spikes and the center DC component represents the frequency of the pattern number of repetitions across the image The higher the frequency the further the spikes are from the DC component and vice versa Therelationship between frequency and distanceis evident in the following sets of images The first image contains 7 cycles vertically and the distance between the spikes and DC component is seven pixels away in the Y direction The second 132 Chapter 6 Advanced image processing image contains 32 cydes horizontally and in comparison tothe first image the distance between the DC component and the spikes is much greater 32 pixels in the X direction The third image exhibits a frequency of 8 cycles in both the X and Y direction Thefourth imageis a combination of theother singlefrequency images and illustrates that the FFT of this complex imageis a combination of the FFTs of the single frequency images As such it alsoillustrates that any image can be broken down and represented by its basic spatial patterns ZH Vertical low frequency image The image in frequency domain Horizontal high frequency image The image in frequency domain Pd Diagonal low frequency image The image in f
39. count will appear in the AllResults table If pills are missing you could usethe coordinates of each pill s centroid to determine the pockets which are empty To familiarize yourself with other blob analysis tools in Inspector run the MissingBlobDemo bas and MoneyToClassify scr scripts included with your software package To do so use the File Open command or Windows Explorer to open the script files The MissingBlobDemo bas Basic script is in the Basic Files folder while the MoneyToClassify scr C Script script is in the Files folder The MissingBlobDemo bas script determines if a pill has been placed in each pocket It uses an alternative approach to the one we just presented it calculates and searches for the pill at each position Then the Inspector script returns a flag each time a blob is not present at the expected position 50 Chapter 2 A few tutorials TheMoneyToClassify scr is an Inspector script that sorts coins in an image counts the different denominations and then returns the values into a text file Example 2 Scaling calibrationand pattem matc hing This second example focuses on processing an image of three pins on a computer chip The image has a square pixel aspect ratio each pixel represents the same real distance both in width and in height but the chip in the image does not have the same size as in the real world since the camera positioning has resulted in a magnified image You will calibra
40. edge might not be found In this case specifying the weights as 3096 for the edge strength and 70 for the contrast will give precedence to the edge with the best match to the specified contrast Inside edges Tohelp find a stripe marker you can specify thetypical number of edges located between theexternal edges of thestripemarker you are defining For example in thefollowingillustration the twostripes sharethesame position sincetheir centers coincide stripe to ignore m expected stripe To find the larger stripe without having to determine and specify its width specify 2 as the number of inside edges of the stripe marker To find the smaller stripe specify 0 this default setting ignores the possibility of any inside edges 220 Chapter 11 Finding and measuring edges and stripes Multiple marker characteristic s Tosearch for a multiple marker enablethe Multi ple checkbox in the Edge and Stripe dialog box Set the edge or stripe marker characteristics just as with any marker except that the expected position and its weight factor are ignored Center of stripe Typical inter stripe spacing EHENENEEENE Measurement box J5M0998 Target image gt Specify the maximum number of edges or stripes to be found in the Number edit field default is 0 meaning that all edges or stripes meeting the established marker criteria in the search box are returned and the typical spacing between the
41. has been rotated or masked these changes are saved with the model Note Inspector can also retrieve model files created using MIL mmo files Fine tuning the model 161 Fine tuning the model Once you have defined your search model you might need to fine tuneit This might involve masking certain regions of the model and or redefining the model s hot spot inthe Dimension tab of the Model dialog box Masking a model Often your search model contains regions that you need Inspector to ignore when searching for the model in the target image These regions might be noise pixels or simply regions that have nothing to do with what you are searching for n these cases parts of the search model can be masked Inspector then ignores the regions under the mask when searching for occurrences of the model Note that the ROI has to be at least as large as the model For example to mask edges in a model perform an edge detection operation on the same region as that from which the model was extracted then binarizetheresulti ng ROI and then use it as a mask as shown below c 5 n RR RE E Original Search Model Defining a mask using the mouse The mouse is usually used to mask noise pixels or pixels close to the boundaries of the model L eft clicking on the mouse will mask the pixel or pixels depending on the size of the brush you can hold down the left mouse button and drag to continously mask Right dick
42. in or out of an image up to 64 times When working with a very small or very largeimage if you perform a processing operation on the zoomed image the zoom factor is maintained in the resulting image Under Windows 95 any zoom factor that would make the vertical or horizontal di mension of theselected image greater than 32767K is not available Under Windows 95 or Windows NT any zoom factor that would make the selected image smaller than 1 pixel by 1 pixel is not available Usethe View Zoom In and View Zoom Out commands to select the zoom factor directly Zoom factors are presented as the proportion of the expected sizeto the original size not to the current size For example a zoom factor of 8 zooms the original image by 8 Zooming occurs from the center of the image Usethe Zoom In Out button when you want to zoom on a specific section directly or are unsure of the required zoom factor When the Zoom In Out button is clicked the cursor changes to a magnifying glass shape and you enter zoom mode Once you arein zoom mode you can zoom in or out on any area of theimage by dicking on the area with theleft or right mouse button respectively Each click zooms the image by an additional factor of 2 in both the horizontal and vertical directions and the selected area is centered when possible When you place your cursor over an image in zoom mode the magnifying glass displays a plus sign if theimageis zoomed in or a minus sign
43. is the match score usually higher than that of the acceptance level above which the algorithm can assume that it has found a very good match and can stop searching the rest of the image for a better one The certainty level is very important because it can greatly affect the speed of the search To understand why you need to know a little about how the search algorithm works Sincea bruteforcecorrelation of theentire model at every point of the image would take several minutes it is not practical Therefore the algorithm has to be intelligent It first performs Setting a model s search constraints 165 a rough sub sampled search to quickly find likely match candidates then checks out these candidates in more detail to see which are acceptable A significant amount of time can be saved if several candidate matches never haveto be examined in detail This can be done by setting a certainty level that is reasonablefor your needs A good level is slightly lower than the expected score If you absolutely must havethe best match in theimage set thelevel to 10096 This would be necessary if for example you expect the target image to contain other patterns that look similar to your model Unwanted patterns might have a high score but this will force the search algorithm to ignore them Symmetrical models fall into this category At certain angles symmetrical models might seem like an occurrence in the target image but if the search was co
44. near the mouse cursor arrow 70 Chapter 2 A few tutorials 6 Right click on the ROI outline and then select Magic Wand Properties from the presented context menu Alternatively right click on the Magic Wand ROI button in the Drawing Tools toolbar and select Properties from the list box presented to obtain the Drawing Tools Properties dialog box 7 Adjust the wand properties to redefine which pixel values areincduded intheROI Set the valueto21usingtheslider or type it directly into the edit field Drawing Tool Properties x Magic Wand Outline Properties for Hi S amp MPLE DCM 0 Selected fi 96 288 103 Similarity R L fi M Greater P Iv Less G j 21 Iv Single object B 21 8 A modified red ROI outlining the vessel is generated 4 Bom Hon ou o mon g bor LR Up Example 3 Working with many images and non standard image file types 71 Adding a fill color to an ROI Ifyou havenot clicked directly on the vessel the defined ROI will not isolate the vessel In this case delete the ROI selected usingthe Edit Delete command or usethe Delete key on your keyboard 9 In addition a Quick View Measured Objects Properties dialog box appears lush Yem bec eed Drip Frogaesstrr Take note of the centroid coordinates X Y for the defined ROI 10 Toverify that the vessel re establishes its initial position in frame 1 play back the sequence Now add a dark fill colour to d
45. or copy them tothe clipboard As with most other operations you can record analysis operations using nspector s scripting tools When you are scripting results from blob analysis or measurement operations can be accessed using the M easGetCur and BlobGetCur script functions which are not recordable Refer to Chapters 16 to 25 for more information on scripting and to theon linehelp for alisting and a description of thesefunctions Result log window 141 Result log window Before performing oneof the advanced analysis operations you should confirm your result log window settings usingtheView Properties command when the Model BLOBSET or measurement table is active or using their respective context menus F rom the Properties tab you can select Log results and Verbose logged results settings If the Log results option is enabled the results of measurements or calculations appear in the Result Log window If the Verbose logged results option is also enabled more detailed results appear in the Result Log window and all results are commented Results and comments can be saved as ASCII text copied tothe clipboard and printed Displaying results You can spedify how to display results of several operations such as blob measurement and pattern matching operations For pattern matching operations you can draw theresult object in the corresponding image To draw the result select Properties fromtheM Model dialog box s co
46. or deselect multiple mode 230 Chapter 12 Defining ROIs and annotating images By default multiple mode is disabled and the current tool returns to select mode once the object has been drawn To temporarily apply multiple mode to a tool double click on it andit will renaininthepressed state until you click on another tool You can also enable disable multiple mode for all the drawingtools by selecting or de selecting Multiple Modefrom thedrawingtools context menu Toset Multiple Modeas your default select the Multiple drawing mode for all tools from the Options Preferences Toolbars tab Selecting a drawing color For annotation drawing tools you can select the drawing color from m Any displayed image using the palette viewer or the eyedropper tool m The Color dialog box Thecurrent drawing color is shown in the color indicators next tothe eyedropper tool The left indicator shows the overlay drawing color and theright indicator shows the image buffer drawing color By default red is the default color in the overlay for RGB images and black is the default color in the overlay for single band images If your image is RGB both indicators will be of the same color If your image is a single band image the image drawing color can only be a color that exists in its display palette therearenorestrictions on theoverlay drawing color Note that the drawing color does not apply to all graphic objects such as ROIs a
47. represent the object and the background 3000 2250 Histogram te 1500 750 Do c7 a 0 64 128 192 255 Automatically determined threshold value 173 A binarization using a g 1 s greater than 173 condition 118 X Chapter 5 Image transformations Edge operations Accentuating edges Many applications highlight locations of an image that are marked by a rapid change in pixel values This is done for example to accentuate the edges surrounding the various objects and features in an image or to highlight defects in a smooth object n general edges can bedistinguished by a sharp change in intensity between adjacent pixels Horizontal edges are created when vertically connected pixels have values that are different from those immediately above or below them m Vertical edges arecreated when horizontally connected pixels have values that are different from those immediately tothe left or right of them m Oblique edges are created from a combination of horizontal and vertical components There aretwo main types of edge operations m One that engances edges to generate higher image contrast m One that extracts detects edges form the image Both these edge operations aretypes of convolutions or spatial filtering operations that replace each pixel with a weighted sum of its neighborhood The weights applied to each pixel in a neighborhood determine the type of operation
48. s view in the case of moving objects I n general this typeof analysisrequires changingthe sourceimagebut using the same set of blob analysis parameters It also involves maintaining statistics on the different sources to determine changes such as position count or size of object s in an image Performing the same blob analysis on different images 185 Performing the same blob analysis on different images nspector allows you to perform the same blob analysis operation on several different images without saving and dosing the existing blob settings and then re loading these settings with a different active image This is useful when you want to perform blob analysis on images grabbed from a camera from a collection or on a grayscale sequence It is also useful when you want to collect blob and measurement statistics from several different images Thefollowingareexamples of when and how you would perform blob analysis on different images m In the simplest case when grabbing images from a camera Inspector s blob analysis operations automatically recognizes if the data in the source image has changed F or an example see cambl ob1 scr n someoperations it might beuseful tocreate a work image You can usethe work image as the sourceimage of your blob analysis operation For an example see cambl ob2 scr m When you have several images open and you want to switch to another source region for your segmentation mask and or g
49. small particles A thick thin operation adds peels layers from an object These operations are similar to the dilate erode operations except if continuously iterated they will not convert all pixels into object pixels as is thecase of a dilation or into background pixels as is the case of an erosion Instead they will eventually reach a steady state known as idempotence In the case of a thin operation this state is usually reached when objects are reduced to their skeleton Thinning until idempotence 124 Chapter 5 Image transformations Note that binary and grayscale thinning to idempotence produce different skeletons For details on the binary and grayscale algorithms see Overview of morphological algorithms If the predefined thick and thin operations do not suit your needs you can create your own structuring element for either of these operations see Creating custom structuring dements Chapter 6 Advanced image processing This chapter describes different advanced image processing techniques 126 Chapter 6 Advanced image processing Advanced image processing nspector contains advanced image processing operations which allow you to remove noise separate objects from their background and correct image distortions They include neighborhood operations using custom structuring elements or kernels and frequency transforms Creating custom filters A spatial filtering operation convolu
50. start up procedure Inspector automatically detects all digitizers in the system that can be allocated and then presents the System Allocation dialog box If your system has several boards of the same type only one board will belisted in the System Allocation dialog nspector will grab with the first board of the specified system type M DEVO If you want tore allocate a digitizer toM DEVI see the readme provided in the online help If your system has several different boards and you want to grab with a different board you must restart I nspector 240 Chapter 13 Grabbing images and specifying camera settings Chapter 14 Creating and manipulating sequences This chapter describes how to create manipulate and grab sequences using Inspector 242 Chapter 14 Creating and manipulating sequences Sequences With Inspector you can grab and playback sequences Sequences are a set of sequential frames that can be either grabbed from a camera built from a number of images or loaded from file of which there are four supported types AVI TIF RAW and DICOM Sequences are useful in applications that Determine the movement growth contraction replication and or metamorphoses of the object s in an image Determine at what interval to grab snapshots when setting up an inspection system by viewing a longer video tape Capture a transient occurrence Reconstruct a large object from a set of images obtained
51. that is Accentuating edges 119 performed F or example certain weights produce a horizontal edge detection whereas others produce a vertical edge detection Edge enhancers Two edge enhancers are available m Sharpen m Sharpen2 m ReliefNW8U Using the Image Processing Filters command you can try both of these edge enhancers to see which is better suited to your application needs Sharpen2 produces more enhanced or sharpened edges After an edge enhancement operation the amplified edges accentuate all objects sothat the eye sees an increase in detail Notethat this operation might not produce good results for further processing because when you enhance edges you also enhance noise pixels Note you obtain approximately the same result as the Sharpenl1 filter by performing a Laplacian edge detection operation on theimage and adding the found edges the original image Edge detection nspector provides a number of edge detection operati ons each offering different advantages depending on your application Usingthe Image Processing Filters command you can select the following edge detection filters from the presented dialog box m Horizontal and vertical edge detection m Laplacian edge detection1 and Laplacian edge detection2 m Sobel1 Sobel2 and Sobel angle m Prewitt 120 Chapter 5 Image transformations Horizontal and vertical Finding the horizontal and vertical edges in an image can be edge detection us
52. the Remove Blobssection This will removeany small artifact from the blob count that has an area less than 100 pixels 48 Chapter 2 A few tutorials 7 Select the Features tab of the Blobset dialog box making raanama imam Canesten Lert ea Bera E 8 Choosethe Centroid features from the Quick Select list box Then click on the Select button Alternatively double click on this feature All centroid features are added for calculation and are listed in the Selected list box Note that a label is always associated to each blob and an area is always calculated 9 Select Count to count the blobs in the image and calculate the specified features minimum area and features related tocentroid A bluelineoutliningeach blobthat has an area above the specified minimum is displayed along with the blob s label number Demos Example 1 Analyzing a blister package 49 In addition an AllResults table is displayed which shows the blob count and lists all blobs meeting specifications induding their calculated blob features 7 Al im Lager Inspector should count the correct number of pills blobs 12 However if an artifact with an area greater than 100 pixels is present in your image take note of its area Then you can return tothe Settings tab and increase the minimum area to a value above that for the remaining artifact s Then dick theCount button again An updated blob
53. the entry but when two or more nodes are added the palette editor linearly interpolates the RGB values of entries between those that have been added as nodes creating a user defined palette Adding nodes is a very efficient way of creating ramps and ranges of color To keep any user defined palette you must save it from the Palette E ditor as a pal file The predefined palettes cannot be changed Palettes can only be saved with the image in formats that support palettes such as tif or bmp but not mim mages opened from such files will usethe palette from thefile instead of the default grayscale palette Seethe on line help Description of Inspector palettes and Creating a user defined palette Creating and using one band color images You can convert a 3 band imageintoa 1 band image with a color palette by using the Image Convert command and selecting 8 bit Unsigned Quantized this command is discussed later in this chapter A 3 band 8 bit color image has 16 7 million possible colors A quantized image has these 16 7 million possible colors reduced to 256 colors in a single band When converting a 3 band image to a single band color image the Octree algorithm is used to reduce the number of color values Octree is one of three algorithms Inspector uses to optimize palettes in an 8 bit display resolution See the section Using palettes with an 8 bit display resolution Color images 85 If you need to process a quantized ima
54. their graphic objects manually Choosing a source and a destination 89 Choosing a source and a destination The ROI on which you apply an operation is called the source region The default source region is the active ROI in the active window The source region will hereafter be referred to as the source For operations that result in an image results are placed into a destination region The default destination region is set using the Options Preferences command in the Image tab The destination region can be either m Theactive ROI m A new image A new image destination inherits the same type as the source if there are two sources of different types the new image destination inherits the greater depth type of thetwo A new image destination also inherits the display options of the source if applicable The size of the new destination region is determined by the operation The destination region will hereafter be referred to as the destination Note that the names of an operation s source and destination regions appear on the Selector at the bottom of the Processing Operations or Geometry Operations dialog box To view the Selector make surethat the View Selector command is enabled Selecting a different source and destination For an operation you can specify a source and or destinati on other than the default one by locking the required image ROI into the required role Locking is accomplished by using the Target
55. this new reference position and takes into account the angular search range of the model 164 Chapter 9 Pattern matching Search time is roughly proportional to the area of the search region If speed is a consideration make the search area as small as possible The acceptance level Thelevel at which the correlation between the model and the pattern in the image is considered a match is called the acceptance level You can set the acceptance level for the specified model in the Search tab of the Model dialog box If the correspondence between the image and the model referred to as the match score is less than this level there is no match The acceptance level affects the number of results returned because only those matches with a score greater than or equal tothe acceptance level will be counted A perfect match is 10096 a typical match is 8096 or higher depending on the image and no match is 0 If your images haveconsiderable noise and or distortion you might haveto set the acceptance level below the default value of 7096 H owever keep in mind that such poor quality images increasethe chance of false matches and will probably increase the search time Note that perfect matches are generally unobtainable because of noise introduced when grabbing images When multiple occurences are found they are returned in the measurement table in decreasing order of match score Setting the certainty level Thecertainty level
56. this will ensure that you keep the number of masked pixels to a minimum Circ ular overscan model search To perform a circular overscan model search follow the same procedure as you would when performing a rotated model search except enable the Use Circular Overscan command When a circular overscan model is created overscan data outside the ROI is also extracted from the source image Specifically Inspector extracts the region enclosed by a circle which circumscribes the model s ROI Model ROI Circular overscan data Consistent central pixels 170 Chapter 9 Pattern matching When a circular overscan model is preprocessed nspector will internally extract different orientations of the model from the overscanned model It is recommended that the model s source ROI be as square as possible the longer the rectangle the smaller the number of consistent central pixels in every model Therefore this type of model should only be used when the model s distinct features liein the center of the region so that they are included in all models when rotated I n addition the pixels surrounding the ROI should be relevant tothe positioning of the pattern that is the model should appear in the target image with the same overscan data or relatively uniform Finally the model must not be extracted from a region too dose tothe edge of the model s source image an error will be generated if circular overs
57. tif image 1 Select the Options Calibration New command The Calibration dialog boxis presented which is used tocreate a calibration object Make surethe Square Pixels option is selected from this Calibration dialog box Note for this example the pixels are square but in nspector you can also create a calibration object that corrects an image with a distorted aspect ratio Assign a specified number of pixels to a specified number of real world units To do so dick on the calibration line which flashes blue when selected Using the mouse draw a line extending from the first edge of the first pin tothe first edge of the second pin in the image The length of this line should be 95 pixels The length will be displayed at the bottom of the image in the status bar Adjust this line if necessary and click the calibration line button a second ti me when finished Inspector places the line s pixel length in thePixels edit field The distance from the first edge of the first pin tothe first edge of the second pin is 2 5 millimeters mm So type 2 5 in the World edit field Select millimeters mm from the U nits drop down list box 52 Chapter 2 A few tutorials Pattern matching 8 Apply the calibration object to the current image by ensuring that the Apply to option is set tothethreepins tif image Then dick OK Note that although the image remains the same after calibration you will now obtain measurement
58. to view statistics for all blobs or only those blobs that belong to a specified class By default results for all blobs are displayed Scroll to view statistics and the corresponding distribution or trend graph for the required feature You can export selected statistics usingthe Edit Export command see the Exporting results section in Chapter 7 for more information To copy selected statistics to the clipboard use the Edit Copy command A trend graph is particularly useful if you want to maintain statistics for a set of images since a trend graph displays statistics in the order in which blobs are found 204 Chapter 10 Analyzing blobs Statistics are also logged in the Result Log window if the Log results option is enabled You can also specify tolerances and view statistics in graphical format by dicking on the Graph button in the StatBI dialog box For more information on options available in the StatBI dialog box See the on line procedure StatBl Dialog Box Options Chapter 11 Finding and measuring edges and stripes This chapter discusses how to find and measure markers with Inspector 206 Chapter 11 Finding and measuring edges and stripes Finding and measuring edges and stripes Many industries base decisions on positi on type measurement of edges or pairs of edges stripes Thel nspector measurement tools can perform a variety of these operations with both a high degree of precision and speed For example
59. types and sizes in one collection Using a collection you can quickly identify the image sequence file to work with from its thumbnail representation or from other information about that file for example its type and size You can also use a collection to group related images that have to be analyzed for a specific application For example a collection could contain images of parts on an assembly line taken with a variety of lighting conditions With these images the collection could also contain images of the same parts showing expected defects This collection could then beused to validate other images with an Inspector script Usethe File New command tocreate new collections Once you createthe collection window you insert imagefilesintoit using the Collection Insert command or by dragging and dropping files from Explorer I mages that are open in the Inspector desktop can also be dragged and dropped intothe collection If you close your collection without saving Inspector will prompt you Inspector will open files according to its extension as long as the extension is listed in theFiles of type drop down list If Inspector does not recognizethefiletype the imageis inserted as an 8 bit grayscale raw image See Raw format earlier in this chapter See the on line help for the procedure To create a new collection Several collections can be opened using the File Open command and then maintained at the sametime however the numb
60. vertical line profiles each of which provides a graph of the pixel values along a small horizontal and vertical line segment respectively to assess the lighting condition in the selected region Then you will perform a histogram of this same selected region since like most images this one contains some noise Then to enhance the image and remove some of the noise you will perform a smoothing spatial filtering operation For image analysis you will segment the pills from the background Since the image has been smoothed you will ensure that the edges of the pills are distinct enough for the Segmentation To dothis you will make use of both thick line profiles Then after segmentation you will perform a pill blob count to confirm the presence of the specified number of pills At the same time you will extract some basic feature information that will allow you to determine which pill is missing if one is missing 1 Load the blispac4 jpg image of the blister package filled with pills from the IMAGES folder Use the File Open command or drag the image from Windows E xplorer into Inspector n general to open more than one image file at a time from thesame folder using the File Open command or Windows Explorer click on the first image file then do one the following a If the files are in consecutive order hold down the Shift key and click on the last item of the group of files to be included Performing a histogr
61. window Now select the Averaged setting from the Continuous tool button s drop down list box This provides the average of 8 consecutive frames Again freezethe grab by clickingtheHalt Grab button or using the File Grab Halt Grab command UseFileSavecommand tosavethis second image grabbed with the Averaged setting as averaged jpg In this part of the example you will perform an arithmetic operation which provides the pixel difference between the normal and frame averaged image 1 Click the Arithmetic operation button Alternatively right click on the image and select Processing from the image s context menu Select Absolute of Difference from thelist box presented in the Arithmetic tab Now you will lock the images into their required roles Source and destination by clicking on the required source or destination image respectivel y Example 4 Grabbing an image 75 Select the first grabbed image normal jpg and lock it as Source 1 To dothis right click on the normal grabbed image and then select Lock Source 1 from the image s context menu Note by default the target Selector bar at the bottom of the Processing Operations dialog box lists the last region activated as Source 1 S1 Select your second grabbed image averaged jpg and lock itas Source2 Todothis select the file AVERGAED_J PG from the Source 2 S2 drop down list box of the Selector bar which appears at the bottom of the Processing Oper
62. 0 around two different centers of rotation In these cases the rotated images have been clipped tothe original image size Original image Rotated 60 Default center of rotation Original image Rotated 60 Different center of rotation Tanslating and changing the symmetry You can physically shift your image using the Image Processing Geometry Translate command You can alsoflip the source image horizontally or vertically using the Image Processing Geometry Symmetry command 114 Chapter 4 Improving image quality Interpolation modes When an image is resized or rotated pixel positions in the destination image get associated with specific points in the source image Then the pixel value for each position in the destination imageis determined from its associated point in the source image and from a specified interpolation mode With Inspector the following interpolation modes are available m Nearest neighbor In this mode Inspector determines the nearest value to a point and copies that value into its associated position In theimage below for example 140 is the nearest value to point 0 5 0 5 so that value is copied into position 2 2 of the destination image pixel position L 0 0 1 0 0 0 1 0 2 0 3 0 pixel value 80 100 80 80 100 100 0 1 1 1 0 1 1 1 2 1 3 1 120 140 80 80 100 100 02 02 22 62 120 120 140 140
63. A Windows display adapter and compatible monitor a A Windows supported mouse or another pointing device In addition to the above you need a Matrox frame grabber in order to grab images Installation You install Inspector by runningthel nspector Setup utility see below Note that Inspector has been developed using the Matrox I maging Library MIL and requires some of its driver and DLL components If you have already installed MIL Lite or if you have installed the full MIL development package Inspector s setup will only re install those drivers and DLLs that are out of date If you have an installed Matrox frame grabber you are presented with the System Allocation list box when you first load Inspector This list box allows you to choose the system you wanttousetorun Inspector It is primarily useful if you have two or more frame grabbers The System Allocation list box is discussed later in this chapter If you abort the Setup utility prematurely Cancel some files might not be copied or modified I n this case you should rerun the Setup utility to ensurethat everything is installed After installation Installation 27 To install Inspector 1 Attach the hardware key to the parallel port of your PC if another device such as a printer is attached to the parallel port disconnect it attach the hardwarekey and then attach the printer connector to the other end of the hardware key If a MIL hardware key is attached
64. Calibration pattern matching take positional and measurement data in uncalibrated pixel units while all tables present results in calibrated units Calibrating your imaging setup Non unity aspect ratio To compensate for a non unity aspect ratio distortion use the Options Calibration New From Grid command and uncheck the Square Pixels option in its dialog This creates a calibration object for which you can specify different pixel to world scalefactors in X and in Y You cantypetheratio of pixels to world units in the edit fields Alternatively you can interactively identify the necessary pixel dimensions by drawing a rectangular graphic object using the Rectangle button in the dialog and typing in the corresponding world dimensions Once you have clicked on this button it will flash to remind you to click on it again after you have finished drawing and adjusting the graphic object Uniform rotation and scaling For rotation distortion or to specify a uniform pixel to world scale you can calibrate your imaging setup using two points in the image wherethe distance between these points is a known value A ruler is often placed in the image so that there is an object of known dimensions Inspector uses the distance between these as a reference for assigning the pixel to world mapping Tospecify a uniform scale usethe Options Calibration New command After ensuring that the Square Pixels option is checked select the Line bu
65. Histograms profiles and measurements You can associate up to 20 profile datasets to a graph window To switch between the different profiles place your cursor anywhere in the white portion of the graph window and click the right mouse button Select Visibility from the context menu then select the graph data set you want to view Note that despite the data set you select to display the grey portion of the graph always displays the data values of the first graph Note that you cannot add profile ROIs to a graph Features of histogram profile graph windows Thereareseveral features common tothehistogram and profile windows Specifically when a histogram or a profile window is active you can Zoom vertically To zoom in out on a graph in the vertical direction usethe View Zoom command Note that when you zoom vertically the entire vertical axis is not visible You can usethe scroll bar to scroll through the vertical axis Alternatively you can have Inspector automatically adjust the size of the graph sothat the maximum vertical value within the two slider bars is visible This is known as interactive zoom mode and can be useful when you are interested in only a range of the histogram profile To access interactive zoom mode select Zoom Interactive from the View menu or from the graph s context menu a Zoom horizontally To zoom in out on a graph in the horizontal direction adjust the width of the graph window a Li
66. Inspector can be used to measure the width of pins protruding from chi ps or printed circuit boards and to measure the distance between each pin Inspector allows you to find sets of image characteristics or markers in animage based on differences in pixel intensities U pon finding a marker the line of the edge s and the edge stripe position are recorded in the measurement table data such as position coordinates angle and contrast are also displayed M easurements can then be taken between results found for different markers or other result objects such as pattern matching positions or graphic objects The measurement tools based on the MIL measurement module rely on a one dimensional imageanalysis As such the measurement tools have several advantages over pattern matching when locating relatively simple image characteristics it is independent of lighting more tolerant of slight differences and much faster You specify the approximate location and other characteristics of a marker to help locatethe marker in an image The more precisely the marker characteristics are defined the more likely itistobedistinguished from similar aspects of theimage The measurement module can operate on 8 bit or 16 bit unsigned grayscale images or on the red component of color images When using 16 bit images all 16 bits should contain image data Inspector will internally convert theimageto 8 bits by dividing by 256 Therefore images contain
67. LOU 0 100 0 Edge value However with the edge at the angle shown below the edge profiles are distributed over ten profiles resulting in a much lower edge strength of 1096 Background Marker pixel pixel value 0 value 255 LoT taeda dy teed Tol Pixel summation ea ea 0 10 10 10 10 10 10 10 10 10 10 0 Edge value 218 Chapter 11 Finding and measuring edges and stripes Determining the strength of the required edge When the edge strength is not very high you should specify the contrast By using the contrast characteristic as a criteria changes in consecutive edge values are summed from where the edge starts and ceases that is from one region of zero edge valuetoanother allowing the marker tobe located n general it is recommended to set the contrast rather than to specify an edge strength since the edge strength is very dependent on lighting To determine the edge strength of the required marker view theEdge Strength graph by dickingthe Graph button in the Edge and Stripe dialog box This displaysthe edge values for every profile value of the search box m Edge Strength Notice that in this example the search box is 80 pixels wide so there are 80 edge values that are returned Thetwo peaks in the graph represent two distinct edges one negative and one positive both with edge strengths of approxi
68. Line Style button on the Drawing Tools toolbar You cannot specify the thickness of the outline of a filled graphic object An outline of one pixel width will be used to calculate the profile of the filled graphic object If you want toobtain the pixel information for filled objects take an X and Y profile X and Y profiles of a region The X and Y profiles sometimes called projections of a region arethesums of pixel values in each column or row of the source region respectively Y Profile X Profile In a grayscaleimage it can be useful to calculatethe X and Y profiles of a region rather than taking a graphic object s outline profile because entire regions can be averaged which minimizes the effect of noise To obtain an X and Y profile of an image use the Image Statistics X Profile or Image Statistics Y Profile respectively Note that the source region need not be rectangular X and Y profile graphs Regenerating the profile window Adding profiles to a graph window Examining pixel values using profiles 151 In X and Y profile graphs the horizontal axis represents the columns or rowsintheregion whilethevertical axis represents the summation of pixel values in each column or row If you are working with a color image threelinesaredrawn on the graph one for each color component Profile window The displayed information in your profile window depends on what is selected from the drop down list box
69. New Blob fromthe image s context menu A blob settings Blobset dialog box appears ji t 1 Click the Blob Analysis button use the Analysis Blob 2 Select the Segmentation tab to identify your objects A segmentation mask is presented 3 Usethe Threshold slider to verify that the background lighting is evenly distributed throughout the entire Segmentation mask image To dothis drag the slider from 70 to 130 Example 1 Analyzing a blister package 47 4 Threshold the segmentation mask using a value that is between themaximum background valueandtheminimum foreground value returned from the previous line profile For this example set the Threshold valueto 115 You can type the setting directly into the edit field or usethe slider 5 Magnify the previewed segmentation mask to view small artifacts To do this select the Zoom button and click on the center of the segmentation mask To increase the magnification continue to left click on the center of the segmentation mask Zoom in 2 more times To zoom out by standard increments right click the mouse Now deselect the zoom tool magnifying glass icon by clicking on the Zoom button again 6 Since there are small artifacts after segmentation which will affect the pill count have Inspector ignore these by choosing a minimum area for the blob analysis search To do this select the Settings tab and type the value 100 in the Min area minimum area edit field located in
70. Ol Onlnputint _OnJump 8 Sna El Enablelnterrupt _ E P DI Disablelnterruptit 12 FindObject ON TurmOnOutputit_ ix FindBlobs OF Turn ffOutputt W AU SetAnalog utputit E Wien ti Al Set ariableT o naloglnput _ ii stema unction PE s GetValue SN eens TIAL SED ti ChangeT able FO FindObject FB FindBlobs SP SetPosition EF ExternalFunction GY GetValue CT ChangeT able I Left Command list with the Vision Command ER 4pc Right Vision Command Group ER 4u SN Snap Captures an image FO Find Object Finds object according to the Pattern Model Name that has been saved in the Pattern folder Returns Number of Object Find Object fis o Enter the name of the model in the Model Name field ViewFlex v 2 6 36 User Manual 0201 Note The name must be the same as it is in the Result Table The SCORBASE variable must be entered into the Number of Object field The results Number of Object will be assigned to the variable All of the robot s coordinates and blob features will be put into the Results Table FB Find Blobs Finds blobs according to the blob analysis name that has been save in the Blobs folder Returns Number of Blobs Find Blobs Enter the name of the model in the Blobs Name field The name must be the same as it is in the Result Table The SCORBASE variable must be entered into th
71. The following is a list of options available in the profile window m f Values is selected the value of the graph at the positi on of these sliders is displayed a f Distance is selected the distance between the values of the graph is displayed This is useful when you need to determine the distance across a blob or the difference between the gray levels of a blob and its background to determine an appropriate threshold The distance can be displayed in calibrated or uncalibrated units depending on whether you select the pixels or world option respectively If Pixels is selected the X and Y pixel values from the uncalibrated image are displayed in the graph If World is selected the X and Y world units from the calibrated image are displayed in the graph If you close the profile window you can have it automatically regenerated for profile objects by selecting the object in the image Adjusting profiles Oncethe profile s graphic object or ROI is drawn it can be moved resized or rotated When you make any of these changes the profile window will be updated accordingly Adjusting the profile object or ROI is useful when comparing pixel intensities of two edges To add profiles to a graph window move resize or rotate an existing profile while pressing the Ctrl key The graph window is updated to display the new profile Therefore all profiles for an image can be associated with a single graph window 152 Chapter 8
72. Toset thefilter options open the Blob Filter Options dialog box by clicking on the Options button from the Blob Edit Filter dialog box You can set the following options a Number of bars Specify the maximum number of bars to use when displaying feature results in the frequency histogram a Number of criteria Specify the maximum number of features that can be used to define a filter Apply after count Specify whether to apply the filter immediately after performing a blob count When this check box is enabled the statistics for excluded blobs are not calculated You can also set this option from the Settings tab of the BLOBSET dialog box or the Settings tab of the Blob Settings Properties dialog box Once you have defined your filter you might want to save it If you save your Blob Settings bst file the filter is automatically saved with these blob settings H owever if you plan to use the samefilter with different blob settings files you can save the filter in its own cls file and then load it into the required Blob Settings dialog Toload a filter from a blob settings file click on the Open button in the Blob Edit Filter dialog box 200 Chapter 10 Analyzing blobs General steps for classifying blobs Classifying blobs With Inspector you can classify blobs that is you can define categories in which to group blobs that have similar specified features You can classify the blobs by defini
73. ViewFlex v 2 8 User Manual 99 Rev D January 2002 intelitek 2002 Intelitek Inc Catalog 100299 Rev D January 2002 Every effort has been made to make this book as complete and accurate as possible However no warranty of suitability purpose or fitness is made or implied Intelitek is not liable or responsible to any person or entity for loss or damage in connection with or stemming from the use of the software hardware and or the information contained in this publication Intelitek bears no responsibility for errors that may appear in this publication and retains the right to make changes to the software hardware and manual without prior notice The printed ViewFlex User Manual that is supplied with the ViewFlex system contains an abridged version of the Matrox Inspector software manual The complete Matrox Inspector User Guide can be viewed and printed from the PDF file named InsptrUG included on the ViewFlex software CD INTELITEK INC 444 East Industrial Park Drive Manchester NH 03109 537 Tel 603 625 8600 Fax 603 625 2137 www Antelitek com ViewFlex v 2 6 3 User Manual 0201 ViewFlex v 2 6 4 User Manual 0201 Contents 1 ViewFlex Installation e eren nnn n neun eee eee 7 Linpacking the EqUIDIBeUL i eo oreet eset to tenta tee ob esso adel dta pe se Lebe aa tetas 7 Installine the ViewElex Systems isnin sie a td a eed etal hia edepncti sistens 7 Assembling the Camera S
74. a good introduction tothe next example topic Example 3 Working with many images and non standard image file types In this example you will open a collection of sample images and work with a sequence of diagnostic medical images This sequence is an angiogram which isolates a major blood vessel and its primary branches during the course of a few cardiac cycles You will play all the frames in this sequence of angiogram images and then do a histogram to obtain a dynamic representation of the pixel values in a defined region in the sequence In addition you will remap the pixel values in some of the sequences grayscale images to confer greater distinction contrast between the blood vessels and the gray background To dothis you will take advantage of I nspector s live preview and locked preview capabilities These will help you choose the Example 3 Working with many images and non standard image file types 63 Loading a collection of images most appropriate remapping After this you will create your own short sequence with images that you will enhance and then apply the selected remapping Finally you will use the magic wand ROI tool to distinguish a specific region of the vasculature in order to show changes in vessel positioning during the circulatory cycle 1 Load the collection of image and sequence files Samplecol from the IMAGES folder To do this use the File Open command or drag the collection file f
75. accuracy but does increase speed Sub pixel accuracy The highest match score occurs at one point and drops off around this peak The exact sub pixel position of the model can be estimated from the match scores found on either side of the peak A surfaceis fitted tothe match scores around the peak and from the equation of the surface the exact peak position is calculated The surface is also used to improve the estimate of the match score itself which should be slightly higher at the true peak position than the actual measured value at the nearest whole pixel The actual accuracy that can be obtained depends on several factors including the noise in the image and the particular pattern of the model However if these factors areignored the absolute limit on accuracy imposed by the algorithm itself and by the number of bits and precision used to hold the correlation result is about 0 025 pixels This is the worst case error measured in X or Y when an image is artificially shifted by fractions of a pixel In a real application accuracy better than 0 05 pixels is achieved for low noise images These numbers apply if you select high search accuracy using the Accuracy drop down list box in the Search tab of the Model dialog box in which case the search always proceeds to resolution level O The pattern matching algorithm for advanced users 179 If you select medium accuracy the default the search might stop at resolution level 1
76. ace without creating a new image view the image in component mode activate the required component and then lock the component as a sourceusingthe Targets Lock Sourcel command If you want a new image upon which to perform an operation extract a band using the mage Extract Band command When using the geometry operations you cannot operate on a single component of an image If you need to perform such an operation on a component of the image you will first have to extract a band Duplicating images 91 Duplicating images Duplicating images is more powerful than simply copying images The Duplicate command allows you to optionally duplicate the data objects associated with the image The data objects that can be duplicated are m ROI objects m Graphic objects Blob objects m Measurement objects note that once duplicated measurement objects become moveable and are no longer associated with measurement table entries This command also copies the selection state of those data objects being duplicated If the image or object is associated with a palette or has been calibrated theduplication command respects those associations Performing a colorordata type conversion With Inspector you can convert an image between color and grayscale convert a color image to a different color space or convert a grayscale image to another grayscaletype by using the Image Convert command Theresulting imageis always placed in
77. adio button is selected graphic objects drawn in the overlay are stored separately from the image Therefore any object can be modified moved and deleted without affecting the image or other objects For more information on handling objects in the overlay seethe on line help procedure Adjusting graphic objects By default objects are drawn in the overlay Once you are satisfied with your annotations you can merge the selected objects into the image by using the Annotations Convert command see Converti ng graphic objects or the Annotations Merge to image command Note that immediately after merging objects to an image with the Annotations Merge to image command an undo operation will restore the image buffer but not the graphic objects A physical overlay frame buffer is not required todraw in the overlay Specifying your drawing preferences 229 When the Buffer radio button is selected graphic objects that are drawn become part of the image Once drawn these objects cannot be modified or deleted except through an Edit Undo I n addition when you save your image these objects are saved as well sincethey are part of theimage This mode is mostly used to create test images and segment blobs that is to separate touching objects If you merge color graphic objects into a grayscale image or if you draw directly into a grayscale image the objects become grayscale The grayscale color used is derived from the selected over
78. am on the whole image Converting a color image to grayscale Example 1 Analyzing a blister package 39 n Ifthefiles arenot in consecutiveorder hold down the Ctrl key while clicking on each file to be included in the group Note that when an image file is of a type that does not correspond to one of those presented in the Files of type drop down list box use the Open As drop down list box to specify the file s data type Perform a histogram operation on this color image To do so click on the Histogram button alternatively use the Image 3 Statistics Histogram command The histogram graph displays the three color bands present in the image default ROI Fiat cage a LE z Lam 07 Equo W Note that the histogram for theentireimageis basically smooth despite the fact that thereis some random noise in the image Tosimplify thestatistical analysis of theimage wewill convert the image to an 8 bit unsigned grayscale image After conversion the statistical graphs generated will show only one band instead of three Working with a grayscale image is useful since you will need one to perform blob analysis later on Inspector can only perform blob analysis on 8 bit one band images however if you do not perform this conversion operation manually Inspector will perform it automatically if necessary for any blob analysis operation 40 Chapter 2 A few tutorials Image processing initial approach
79. an 8 bit image by enablingthe Grab Depth check box intheDigitizer Info tab of the Digitizer Settings dialog box An 8 bit grab can also be forced by using the Options Preferences Digitizer tab command and checking the Always default to 8 bit grab on new digitizer installation checkbox Note that unless you havea Matrox Pulsar this box should remain unchecked Configuring the synchronization and signal channels If your digitizer supports simultaneous videoinput connections of the same type you can switch between the input channels The channels can be switched using the Options Digitizer Settings command Inspector identifies the channels accordingtotheir corresponding label on theboard s connecti on plate usethese identifiers when accessing multiple cameras Inspector names the channel indentifiers from O Certain digitizers have separatesignal and sync channels and therefore have a channel reserved for either signal or sync input If you areresetting the input channels and see one missing from the drop down list it is because that particular channel is reserved To determine which channel inputs are available on your board consult your board s installation manual or see the on line help topics Input channels on Corona Input channels on Genesis Input channels on Meteor II Input channels on Meteor and Input channels on Pulsar Using multiple digitizers 239 Using multiple digitizers By default during I nspector s
80. an image 73 Chapter 3 Dealing with images 77 Images in general eee 78 Loading an image eese 79 Grayscale or single band images 80 Signed images lees 80 Processing a grayscale image 81 NonS8 bitimages elles 81 Packed binary images 81 Displaying grayscale images 81 Displaying single band images with a palettes m oi a suatia i ale aa ed 83 Color images esras nia eyes aire are EUEN 85 Multiple views llle 85 Color Spaces ng seb hha ae ae mn 86 Representing color images 86 Operating on regions of interest inyourimage eee 87 Defining ROIs 0 08 87 ROIproperties llle 88 Choosing a source and a destination 89 Selecting a different source and destination ores rinrin ee rea nea 89 Processing color images 90 Duplicating images aaee 91 Performing a color or data type conversion 91 Other image and sequence file types 93 Rawformat leeren 93 DICOM medical format 94 Collection files llle 95 ZOOIDIng oua x AN Cer UV MUS SN 97 Eram SIZE Pe aw Rae 98 Accessing context menus in zoom mode 98 Special scrolling mode 98 View AS Lexi sehr p Gis tan RS RA ENTRAR GR 98 View 3D surfaces leere Using palettes with an 8 bit displa
81. and measurement a Click Set Distance m Click Next Step This will take you to the Step 2 dialog box 28 User Manual Calibration Click on the pen button essa 09 8849 12241 72 352 X 288 Calibration Step 2 dialog box 9 Ifnecessary move robot arm out of the camera s view 10 Click Snap in the Step 2 dialog box 11 You will be able to recognize Pos 1 and Pos 2 in the image Click the Pen icon The Pen icon will start flashing after being engaged Your mouse arrow will become a cross Click on Pos 1 then make a dragging motion using your mouse toward Pos 2 This movement represents an increase in the X axis shows the direction of increase of the X axis 12 Click the Pen icon again Now the X and Y axes are set and the distances are in real world millimeters When you move your mouse along the image Pos 1 becomes coordinate 0 0 center point and the directions of X and Y are now the same as the robot The movement of the mouse along the axes shows changes in X and Y 13 Click OK Now the Vision System is calibrated and synchronized with the robot and all new images will become calibrated Opening and Saving Calibration Files ViewFlex v 2 6 0201 The Configuration Step 2 dialog box enables file management operations Save The Save icon saves the calibration The extension of the files is CAL Open The Open icon will allow you to load backup calibration files When the OK icon
82. art and end anglein a sourceimage The function scans the specified zone from the start angletothe end angle n our example since the start angleis less than the end angle the direction of the scan is counter clockwise The height of the destination image is equal to Start radius End radius The width of the destination image is determined by the arc length The arc length in pixels is equal to outer radius Start angle End angle where the angles are in radians The valid range of angleis from 360 to 360 degrees and the maximum span of theanglemust not exceed 360 degrees These values are then mapped to a destination image A polar to rectangular transform performs the reverse of the transform described above It takes a sourceimage and maps it to a destination image The center start angle end angle start radius and end radius fields are used to specify the position of the contents of the source image in the destination image 138 Chapter 6 Advanced image processing Watershed transformations You can separate touching blobs by checking the Separate check box in the Segmentation tab of the Blob New command Blobs are separated when a count operation is performed The segmentation mask displays a preview of of a watershed transformation in an image A watershed algorithm is used to separate touching blobs Inspector provides a Watershed exe utility script to perform watershed transformations It is locat
83. ary in some cases thetimerequired tore paint the display reduces the maximum bandwidth available for the actual grabbing of data After you call the File Grab Sequence command you are presented with the Sequence Setting dialog box with the Timed Acquisition Setup tab showing In theTiming section of the tab you can select to use the digitizer s timing which will grab every frame or you can select tograb every nth frame or every n milliseconds Thesequenceis grabbed only when the Start button is clicked if you select the Stop grab at user request only checkbox in theTimed Acquisition Setup tab the digitizer will grabinto acircular queueuntil theStop button is clicked When the stop grabbing command is under your control the grab is known as grabbing intoa circular queue Thegrab can be paused by using the Pause button To optimize your system s performance free up memory by closing all unnecessary images prior to grabbing a sequence See the on line procedure To grab a sequence Grabbing sequences 247 Grabbing memory type sequences You can grab a memory type sequence to paged H ost memory system RAM or on board memory or DMA memory non paged Host memory The memory available for the grab determines the maximum length of the sequence the more memory available the longer the maximum sequence length The grab destination determines the speed of the grab as well as the memory available for the grab The speed of t
84. ath a cabinet Image Sharpness Drag the slider left to soften the image drag the slider right to sharpen the image USB Bandwidth When you have multiple USB devices in use at the same time and the camera is using too much of the USB bus use this control to specify how much of the bus to allocate for the camera Drag the slider to the left to decrease the amount of the bus used by the camera drag the slider to the right to increase the amount of the bus used by the camera Changes made here won t take effect until video is stopped and restarted CPU Usage When using the camera on a computer with a processor running at 233MHz or less the minimize CPU Usage checkbox is enabled by default On computers with faster processors only use this control when the computer becomes sluggish while using the camera Click Restore Defaults for All Camera Settings to restore all of the tab settings to their factory settings 18 User Manual Pan Tilt Zoom ViewFlex v 2 6 0201 Controls the location of the image in the field of view Video Source 2 x Main Exposure Color Image Advanced Pan Tilt Zoom a a Zoomin Zoom Out s iF Pan Tilt Zoom tab Pan Moves across the image horizontally This option does not work when the camera is Zoomed all the way out or when the capture size is 640 x 480 Tilt Moves across the image vertically This option does not work when the camera is Zoomed all the w
85. ation project files 28 Chapter I Getting started Uninstall under Windows 95 98 or NT 4 0 Notethat if you add a frame grabber after installing nspector you should uninstall MIL from the Control Panel otherwise the drivers will not be installed Then rerun the Inspector Setup utility This will ensurethat the necessary files are added You can uninstall Inspector as follows 3 Select Settings from the Start menu 4 Select Control Panel from the presented sub menu 5 Select Add Remove Programs from the Control Panel menu The Add Remove Programs Properties dialog box appears with the Install U ninstall tab active 6 Select Matrox Inspector 3 0 from the presented list box 7 Click on Add Remove To completely removel nspector from your system you should also uninstall Matrox Imaging Library Products if present System Allocation 29 System Allocation If you have an installed Matrox frame grabber you are presented with the System Allocation list box when you run Inspector This list box contains all the systems that I nspector detects on your computer If your computer only has one board you can check the Don t ask me again option and the next time you open Inspector it will open using that board The system choice you make is saved in the registry If you arealways going to use the same system torun Inspector you can disable the System Allocation list box for future sessions in which case the system sp
86. ations dialog box The list box presented lists all images that can be designated as S2 Alternatively you can lock the grabbed frame averaged image by right clicking on the image and then selecting Lock Source 2 from the image s context menu Lock the destination image D as a new 8 bit unsigned image Todothis chooseusetheSelector bar at the bottom of the Processing Operations dialog box Now dick on the Apply button to implement this arithmetic operation Note to remove the locks on all targets without performingthe operation thatis toreturntothe default source and destination images select Unlock AII Targets from the image s context menu Select the Mapping tab and click on the Window Leveling operation button to remap the pixel value range toa full range 76 Chapter 2 A few tutorials Chapter 3 Dealing with images This chapter describes the types of images Inspector supports dealing with grayscale single band and color images image conversion image duplication and band extraction It also discusses how to use ROIs special file types how to use collection files and special display effects such as zooming view as text view 3D and using palettes 78 Chapter 3 Dealing with images Images in general Inspector supports the loading saving and creation of grayscale and color images as well asthe grabbing displaying directly or indirectly and processing of these image types Once you h
87. atures In addition they allow you to take a variety of measurements between their positional results and get quick on the spot measurements such as length area and angle For most imageanalysis operations you can maintain statistics in Inspector Typical application steps Although there are many applications for which Inspector can be used most involve the same basic steps 1 2 3 Load an image from disk or grab one from an input device To optimize the image s use nspector s enhancement and transformation operations You might need to use some of Inspector s analysis operations to determine your optimization strategy Togeneratethe required imageinformation usel nspector s analysis operations Save and or export numeric results Inspector can export results to most popular spreadsheet and database packages Calculated data can then be compared with existing data or used in statistical computations A few examples 37 A few examples To get you started the remainder of this chapter will serve as a tutorial As such we provide four examples that will familiarize both novices and advanced users with Inspector s operations The four examples have been chosen to introduce you to processing operations as well as reveal shortcuts suggest hints and alternative methods for efficient use of nspector s imaging software E ach example is presented as a sequence of operations and lists the steps to
88. aulty parts by comparing calculated blob features with a set of expected feature values This generally involves using blob filters statistics and tolerances Tissue and material analysis The purpose of this type of blob analysis is to analyze complex tissue Such as biological tissue to determine any of the following existence of certain types of blobs number of blobs in an image or distribution of the blobs The image used in this type of application is usually very difficult to segment and thus requires sophisticated segmentation techniques F or example the image might contain touching objects that must be separated before they can be counted accurately Therefore it might be useful to Segment a specific region of interest ROI of the image import the mask into blob analysis extract the required information and then changethe sourceimage and mask to another region of interest in the image Toavoid countingthesameblobstwice when movingtheROI you can disable counting blobs that touch two of the ROI border edges F or example if you are moving the ROI across the image in a left down pattern you can disable counting blobs that touch the top and left edges to avoid double counting a blob Sequence analysis Thepurposeof this typeof blob analysis is to study a set of images acquired over a known period of time During this period of time one or more factors might be variable for example the number of objects in the camera
89. ave loaded grabbed or created an image it can be zoomed converted into another data type or associated with a display palette You can perform operations on an entireimage individual color components of an image or on a localized region of interest Several images can be grouped into collections to facilitate maintaining images Inspector supports images with the following data types and pixel depths m 1 bit packed binary images m 8 bit 16 bit and 32 bit signed or unsigned images m 8 bit 3 band RGB and HSL images m 32 bit floating point images Inspector supports loading images from single image files for example TIFF BMP and J PEG sequence files for example AVI and TIFF and raw and DICOM images For a list of supported file types see Appendix B Notethat 4 bit images can beloaded but they are converted to 8 bit In addition HSL images can beloaded or saved but color space information is not retained See Color images With Inspector you can create a color or grayscale single band image and fill it with a specified value See the on line procedure To create a new grayscale or color image Loading an image 79 Loading an image An image can be loaded in one of several ways most of which are standard Windows methods of opening files a The File Open command m lIheRecentFilesand RecentScriptslistin theFilemenu m A collection file discussed later in this chapter m Explorer Select the require
90. ay out or when the capture size is 640 x 480 Zoom In Zooms in on the portion of the image currently visible in the video window The size of the rectangle shows the amount that the camera is zoomed Zoom Out Zooms out on the portion of the image currently visible in the video window The size of the rectangle shows the amount that the camera is zoomed Shortcuts Right clicking the mouse inside the pan area opens a menu with the following e MaxZoom In the camera zooms all the way in e MaxZoom Out the camera zooms all the way out e Center View places the image in the center of the field of view e Always on top prevents the Camera window from being hidden by other windows 19 User Manual Camera Connection Client Camera only Connect to If there is more than one ServeCam on the network then you can make a connection to the specific ServeCam that you want by entering the IP address of that ServeCam s computer The default IP address is the one that initially appears in the box which was automatically detected Clicking OK will connect to the ServeCam at the IP address shown ACL Terminal for ViewFlex ViewFlex v 2 6 0201 il ia V EDT ACL Advanced Control Language is an advanced multi tasking robotic programming language The ACL Terminal for ViewFlex is the software interface that provides access to the controller from a PC and provides functions needed to configure program and operate the ro
91. ayscale values greater than or equal to 1 are changed to 255 The region from which to create the mask must be the same size as the existing source region 194 Chapter 10 Analyzing blobs To import a segmentation mask make sure that the source region from which to create the mask is active Then open the Blob Settings Properties dialog box one of the following Ways m Clickingon the Mask button from the Segmentation tab or the Properties button from the Settings tab of the BLOBSET dialog box a ByusingtheBLOBSET dialog box or All Results window s context menu and selecting Properties from the displayed menu a When the BLOBSET dialog box or All Results window is active select Properties from the View menu In the Mask tab select thelast active ROI option The data from this ROI is used as the permanent mask Specifying your segmentation preferences With Inspector you can control how certain aspects of your segmentation mask are interpreted You can control a Which pixel value 0 or 255 is considered a foreground blob pixel Foreground options in the Settings tab m Theminimum and maximum area of a blob Min area and Max area options of the Settings tab Blobs whose areas are not in this range are excluded from calculation m Whether to exclude from calculation either automatically or explicitly blobs touching the segmentation mask borders Touching options of Settings tab m Whether to exclude from
92. bject orientation is required the shape and distinctiveness of the object the complexity of the image background and the degree of angular accuracy that is required With the Analysis Pattern Find Image Orientation command you can quickly determine the orientation of an image based on the dominant edges in the image and their angular displacement from the image frame The image can have either uni directional dominant edges such as parallel stripes or bi directional perpendicular dominant edges The operation is designed for images with smooth edges usually obtained when grabbing an image with a camera It will not work well on an artificially generated image unless the lines and edges are anti aliased 30 degrees 30 degrees typical image edge detection on a 30 degree rotated image Oncethe orientation of a target image is found the angle is returned to measurement table and an angular search can be performed within a narrow range of the angle 172 Chapter 9 Pattern matching Model orientation Note that if an image does not have dominant edges its orientation might not be found properly In addition if the image s background contains perpendicular edges the orientation of these edges might be found instead Theorientation of a singlelarge object on a smooth uniform background can be found using the Analysis Pattern Find Model Orientation command This searches for the general contours of the object in a target
93. bot system Fs ACL Terminal for ViewFlex B x File Communication Robot lial sj exp Ale o zl Status Settings 9600 n 8 1 00 00 33 4 ACL Terminal for ViewFlex Save Saves a listing of all user programs variables and positions in one file The files are in ACL format acl Download Downloads data from a user backup file in the host computer to the controller Does not erase or modify existing programs Abort All Immediately aborts all running programs and stops movement of axes On Line Off Line Opens closes the communications port 20 User Manual AaB m E ON CON Enables servo control for all axes a specified group of axes or a single axis OFF COFF Disables servo control for all axes a specified group of axes or a single axis Home Drives all robot axes to their home position Port Properties Opens a dialog box that allows you to configure the RS232 serial communication port in the PC that communicates with the robot controller Properties Pat co m Maximum Speed Echo s rj Off amp On Cancel r Connection Preferences Data Bits a Parity Noe Stop Bits ae aan ACL Terminal CommPort Properties ACL ViewFlex String Commands The ViewFlex String Commands are written as Print Printin ViewFlex string commands ViewFlex helps you make applications that combine vision and robot guiding that will use the ACL lang
94. bs and or their position in the image which should belong to the class without knowing the actual combination and values of the features for these blobs Todefinea training set class you must select one or more blobs whose selected features ultimately will be used to define that class 202 Chapter 10 Analyzing blobs Saving and loading classes Weight of a feature Theblobs included in theclass depends on theselected measure of similarity set using tolerances or specified ranges To set similarity refer to the on line procedure Setting classification options Once the required blobs are selected the statistics for all features are extracted from the blob count results If the standard deviation for a feature is small this feature might be a significant one to define the class since the selected blobs have similar values for this feature You can use any I nspector feature s todefinea class training set n other words features used to definea class donot have to be enabled as one of the features for calculation in the Features tab of the BLOBSET dialog box For more information see the on line procedure Training se method Once you have defined one or more classes you can savethem If you save your Blob Settings file bst theclasses are saved with these blob settings H owever if you plan to usethe same dass settings with different Blob Settings files you can save theclasses in its own file cl s a
95. but their appearance can be improved by redrawing them using a matching schemethat fi ndstheclosest colors in thehardware palette for each of the image s requested colors Since repainting of the background images takes ti me you can choose not to repaint them if their display quality is not important a Don t care about Background Quality When the Own Palette option is enabled the palette manager ignores the palette requirement for theimage s background E ssentially this option when selected disables the background image remapping 102 Chapter 3 Dealing with images Palette optimization ThePalette Optimization command is only available in an 8 bit display resolution when theOwn palette option has been selected from the Options Preferences command and when the current images a 3 band color image This command remaps the pixel values to 256 discrete values In effect thePalette Optimization command improves the appearance of the imageon thedisplay Note however that the command requires a fair amount of computation therefore it is not performed automatically whenever the content of a 3 band image is processed ThePalette Optimization command is most useful if your imagehas a largerange of colors that do not exist in the common palette or if you want a high quality grayscaleimage Inspector uses three industry standard algorithms to optimize the palette Median cut Popularity and Octree You can select one of these algor
96. by panning across the object with a camera Sequences can be in one of two forms Memory sequence Theframes of this type of sequenceare stored in your system s memory That is each framein the sequenceis allocated its own memory image buffer You can process and analyzetheframes in a memory sequence as you would any other i magein nspector Oncea memory sequence is saved it becomes a disk sequence Disk sequence The frames of this type of sequence are savedto disk and cannot berearranged or changed Only one image buffer is allocated and used for the display of any frame You can create an image from any frame in the sequence as well as drag frames into a new or existing memory sequence Loading sequences and the sequence window 243 Loading sequences and the sequence window You can load a sequence file using the File Open command or from a collection window Sequence files can be of type AVI multi page TIF DI COM or RAW Sequences are displayed in a window which has two panes a main playback pane and an optional thumbnail pane Either of these panes can be closed or made smaller by dragging the vertical edge which separates the two panes If the sequence you are opening contains more than 150 frames you will be asked if you want to display the thumbnails While a sequence is loading you can press the Escape key to stop creating the thumbnails The sequence window displays the name of the sequence the current fra
97. calculation blobs whose specified features are outside a specified range refer to the Filtering blobs based on features section m Whether two diagonally adjacent pixels are considered touching Settings tab of the Blob Settings Properties dialog accessed using the Properties button In Inspector horizontally and vertically adjacent pixels are considered Specifying your segmentation preferences 195 touching You can specify whether diagonally adjacent pixels are considered touching by selecting 8 neighbors in the Settings tab ry d x N a e v v If diagonally adjacent pixels If diagonally adjacent pixels t are not considered touching are considered touching a blob has 4 neighbor a blob has 8 neighbor connections connections For example the following is considered one blob if diagonally adjacent pixels are considered touching but is considered two blobs if they are not 196 Chapter 10 Analyzing blobs Sorting results Counting and calculating selected blob features Onceyou havesegmented your imageand specified thefeatures tocalculate you can count the blobs in theimage and calculate their features To do so click on the Count button from the Settings Features or Segmentation tab of the BLOBSET dialog box Alternatively select Count from the BLOBSET dialog box s context menu The results of all blob calculations are returned in a re
98. can data extends beyond the boundaries of the image Searc hing fora model within a range of angles When an angular search range has been specified the preprocess operation creates a model for every x degrees within the range where x is specified by the Tolerance After the approximate location is found Inspector fine tunes the search accordingtoa specified Accuracy To beeffecti ve you must set the degree of accuracy to a value smaller than that of the rotation tolerance When searching within a range of angles you should use as narrow a range of angles as possible since the operation can take a long time to perform Tolerance defines the full range of degrees within which the pattern in the target image can be rotated from a model at a specific angle and still meet the acceptance level When creatingtheinternal models nspector rotates according tothe specified interpolation mode These settings can influencethe speed of the preprocessing operation significantly The accuracy of the search can also be influenced Once a search has started it cannot be stopped by simply pressing on Esc you must shut down Inspector Image orientation Quickly finding the orientation of an image or model 171 Quickly finding the orientation ofan image or model You can find the angular displacement of a target image or of an object in that image in a number of ways The choice of method will depend on whether whole image or o
99. cedure Continuous grab Setting digitizer controls During a continuous grab you might be presented with the Digitizer Control dialog box This dialog box allows you to interactively adjust the various dynamic controls of your digitizer Use the Option Preferences command to specify whether this dialog box should present only standard options or both standard and advanced options The standard options appear in the References tab of the Digitizer Control dialog box The advanced options appear in other tabs that vary according to your particular frame grabber Standard options The standard options allow you to adjust the black and white reference levels of your frame grabber s analog to digital converters or the hue and saturation of its composite decoder By reducing or increasing either or both the black and white reference levels you can affect the brightness of the resulting image By reducingonereferencelevel and increasingtheother you can affect the contrast of the resulting image The hue and Grabbing 237 saturation should normally not need adjustment unless you need to calibrate to a color standard one or two points should be sufficient Note that the standard options are recordable when scripting For example the black and white reference levels can be adjusted through the scripting function CamSetRef In addition the standard options are saved in the Inspector GrabControl page of the Windows registry Advanced opti
100. ch for the marker Proper placement of the search box is essential to the success of any Find operation The search box should belimited to as small an area containing the marker as possible in order to ensure the success of the operation especially when using a highly detailed or complex target image In addition by limiting the processing region you can accelerate the find marker operation Thesearch box can beplaced manually on theimageby clicking on the Draw Search Box button in the dialog box of the Analysis Measurement Edge and Stripe command place thecursor in therequired location left click and dragthesearch boxtoits appropriate sizeand click on New If you donot want to draw the box click on New and a default search box will be drawn using the center of the image as the origin You must adjust the search box s Search search box direction Location pst line m Size and position to specify the area in which to search m Angleto ensurethe proper search direction Note that the search takes place in the same direction as the search direction line m Adjust the search box s location line to indicate the expected position of the edge or stripe optional 210 Chapter 11 Finding and measuring edges and stripes Obtaining valid results To obtain valid results the edge or stripe must enter and leave by opposite sides of the box Theillustration below is an example of valid and invalid search box definitions Stripe
101. ch time against the increased chance of not finding the pattern at very low resolution Therefore it chooses a starting level according to the size of the model and the characteristics of the pattern In the application described earlier 128x128 model and 512x512 image it might start the search at level 4 which would mean using an 8x8 version of the model and a 32x32 version of the target image You can if required force a specific starting level or last level using the Search levels section of the Model dialog box s Advanced tab Thelogic of a hierarchical search accounts for a seemingly counter intuitive characteristicof the pattern matching search operation large models tend tobe found faster than small ones This is because a small model cannot be sub sampled to a large extent without losing all detail Therefore the search must begin at fairly high resolution low level where the relatively largesearch region results in alonger search time Thus small models can only befound quickly in fairly small search regions Search Heuristics Even though performed at very low resolution the initial search still accounts for most of the computation time if the correlation is performed at every pixel in thesearch region On most models match peaks pixel locations wherethe surrounding imageis most similar tothe model and correlati on results are largest are several pixels wide These can be found without evaluating the correlation funct
102. check boxes You must specify thesize of theimageframe nspector guesses the image s dimensions based on thefile s size displayed at the bottom of the Raw Image Information window and the pixel depth which is selected from the radio buttons The size of a given frame of raw data is defined in bytes and the width ti mes the height is the number of pixels in the image Frame size bytes pixel width height If the settings which appear for your image are inappropriate start at the Pixel depth radio buttons and the image s height and width areautomatically recalculated If you open thei mage with inappropriate settings you will not be ableto work with the image 94 Chapter 3 Dealing with images Iftheraw data is a sequence the Number of frames field will show thenumber of frames in the sequence which is calculated based on the specified size of the frames DICOM medical format DICOM is an industry standard format for representing and transmitting medical images and was developed as a method of communication between a medical instrument and the computer Typical types of DI COM images are a Ultrasound type images Theseareusually low resolution grayscale images with 256 intensities Typically these images are in sequences which render more information to the user a Cardiac based images These are usually of medium resolution containing up to 10 bits deep Typically these images are in sequences which render m
103. complete the relevant task at hand Thefirst and second examples serve as a warm up They allow you to perform some general commands for example load an imagefilefrom disk as well as enhanceand analyzean image Sinceblob object analysis and pattern matching recognition are the most common uses of I nspector s image processing capabilities thefirst and second exampl efocus on blob analysis and pattern matching respectively The second example has been chosen tointroduce you tol nspector s calibration and edge stripe finding capabilities Thethird and fourth examples focus on dealing with many images and grabbing digitizing an imagefrom an input device camera respectively These examples build on techniques presented in the first two examples and allow you to take full advantage of useful commands and or operations such as window leveling and locking targets 38 Chapter 2 A few tutorials Loading an image file into Inspector s workspace hi Example 1 Analyzing a blister package In the first example you will analyze an image of a blister package that could befound along a pharmaceutical inspecti on line For the initial assessment you will perform a basic histogram operation which gives you a intensity distribution graph of the entire image Then you will focus on a selected area in an image to determine an enhancement and analysis strategy suited to the image You will perform thick horizontal and
104. ctors This is evident for example if you know that the object in your image should be round and it appears as an ellipse This type of distortion is often a side effect of the sampling rate used by some older digitizers Rotation distortion Present when the camera is perpendicular to the object grabbed in the image but not aligned with the object s axes a Perspective distortion Present when the camera is not perpendicular totheobject grabbed in theimage Objects that are further away from the camera appear proportionally smaller than the same size objects closer to the camera a Other spatial distortions Complex distortions such as pin cushion and fish eye distortions fall in this category These distortions can be compensated for by using a large number of small sections in the mapping function If the number of Sections used is big enough and the corresponding area covered in each is small enough the mapping in each area can be approximated with a piecewise linear interpolation function Calibrating in Inspector 255 Calibrating in Inspector Inspector provides two ways to calibrate your imaging setup m The Options Calibration New command allows for the creation of a calibration object that compensates for non unity aspect ratio and rotation distortions m TheOptions Calibration New from Grid command allows for the creation of a calibration object that compensates for perspective and other spatial distortions Fo
105. cussed later in this chapter 86 Chapter 3 Dealing with images Color spaces The color space of a color imagerefers to the way color in that image is represented Common color spaces are RGB and HSL RGB colorspace In the RGB color space color is represented as a combination of red green and bluecomponents This color spaceis generally used for display hardware since it best matches the three colored phosphors of display monitors It is alsothe color space of many cameras and input devices HSL color space In the HSL color space color is represented as a combination of hue saturation and luminance components The hue component can be thought of as the actual color information The 360 degrees of the ideal color wheel are mapped onto 256 values starting with red at 0 and going through yellow green cyan blue and magenta Saturation can bethought of as the measure of color purity or concentration Luminance corresponds tothe brightness of the colors and is what a grayscale version of the image looks like TheHSL color spaceis similar tothe human way of describing colors Each color has its own hue value such as red orange or green Once the hue value is chosen changes to the saturation or theluminancealter only the color quality not the basic color Representing color images RGB and HSL true color images are represented using three 8 bit bands of data where each band of data represents a color component of
106. cy components There are two main types of spatial filters which can remove high frequency components low pass filters and rank filters Low pass spatial filters are effective in reducing Gaussian random noise and high frequency systematic noise when the noise frequency is not too close to the spatial frequency of significant image data These filters replace each pixel with a weighted sum of each pixel s neighborhood Note that as these filters remove noise they havethe side effect of selectively smoothing your image and reducing edge information Inspector provides several types of pre defined low pass filter operations among them Average Smooth and Smooth 5x5 Use the Image Processing Filters command to select one of these filters The average filter places the same importance on all pixels in the neighborhood whereas the smooth filters place more importance on the pixels that are closer to the center of the neighborhood All the predefined filters operate on a 3x3 neighborhood except the Smooth 5x5filter which operates on a 5x5 neighborhood Rank filter operations are more suitable for removing salt and pepper type noise sincethey replace each pixel with a value from its neighborhood rather than with a weighted sum of its neighborhood A median filter is a type of rank filter that replaces each pixel with the median of its neighborhood nspector provides a pre defined median filter which operates on each pixel s 3x3 neighborho
107. d a filter kernel in the spatial domain corresponds toa point to point multiplication between the image and filter in the frequency domain M ask locations with the value 1let corresponding frequencies pass and mask locations with the value 0 block or remove unwanted frequencies For aFourier transform whose DC component is in the center a low pass frequency filter smooth is a circle filled with the value 1 centered on the image and a radius that corresponds to the highest frequency you want to let pass Similarly a band pass frequency filter is a torus donut When you want to remove a particular frequency spike it is important that you also remove the corresponding negative mirror spike which is symmetricto the origin If you do not remove the corresponding spike there transformed image will contain ghost artifacts The possibility of ghost artifacts and the difficulty in drawing very small ROI s accurately by hand suggests that frequency filters should be created with a script rather than manually Inspector provides two utility scripts for creating frequency filter masks FFT operation bas written in Inspector Basic and FFT exe which gives a slightly more sophisticated dialog for constructing filter masks The masks that are constructed can besaved using theFile Save command and can be applied as filters in later sessions of Inspector FFT exe which is located in the Util subdirectory of your Inspector installation co
108. d image s and then drag them into the Inspector workspace All methods except for the Recent Files and Recent Scripts methods allow you to open several images at once You can change the maximum number of most recently used files shown as well as choose whether to display the complete path of each file or just the names using the Options Preferences command in the General tab You can view information about an image before loading it using theI nfo button of the File Open command In addition you can obtain information about the image such as file type and size after it is loaded by using the View Properties command Inspector will open files according to its extension as long as the extension is listed in the Files of type drop down list If your image has an unfamiliar extension 7 you should specify how the image should be loaded by selecting the appropriate item in the Open As field of the File Open command F or example you can specify to open an imageas a raw imageusing this field If Inspector cannot determine the file type the image will be opened as a raw image See the section Raw format 80 Chapter 3 Dealing with images Grayscale orsingle band images A single band image is an image whose data has two dimensions X and Y These images are usually monochrome or grayscale grayscale images are essentially a subset of single band images Single band images however are not exclusively grayscale the
109. display method linearly maps the values in theimage sothat the minimum and maximum values in thedisplay buffer are set to 0 and 255 respectively Autoscaleis the default remapping mode Note that if the image contains a single value that is if the image was created and filled with a single value this value is mapped to 0 127 or 255 depending in which third of the total data range in the image this single value falls a Bitshift The bitshift display method bitshifts the image data sothat the 8 most significant valid bits of data are displayed F or this method you should specify the number of bits that actually have significant data F or example if you are grabbing into a 16 bit buffer with a 10 bit camera use the bitshift display method and set the significant bits to 10 This method should be used when you know how many bits of the image contain valid image data for example when storing a 10 bit image data in a 16 bit image This ensures that as much of the available dynamic range of the 8 bit display buffer is used Thisis particularly useful when most but not all of your image values lie within a small range When selecting the bitshift method you can also specify the number of bits that contain valid image data a Logarithmic The logarithmic display method is a logarithmicremapping of values which emphasizesthelower values in the data range This is particularly useful when your image is a 32 bit floating point ima
110. e model are produced and the search begins on a much reduced scale This series of sub sampled images is sometimes called a resolution pyramid because of the shape formed when the different resolution levels are stacked on top of each other Each level of the pyramid is half the size of the previous one and is produced by applying a low pass filter before sub sampling If the resolution of an image or model is 512x512 at level 0 then at level 1it is 256x256 at level 2 itis 128x128 and so on Therefore the higher the level in the pyramid the lower the resolution of the image and model Level Resolution Search stage 7 Lowest Initial 6 5 l 4 3 2 1 O Highest Final Thesearch starts at low resolution to quickly find likely match candidates It proceeds to higher and higher resolutions to refine the positional accuracy and make sure that the matches found at low resolution actually were occurrences of the model Because the position is already known from the previous level to within a pixel or so the correlation function need be evaluated only at a very small number of locations Since each higher level in the pyramid reduces the number of computations by a factor of 16 it is usually desirable to start at as high a level as possible However the search algorithm The pattern matching algorithm for advanced users 177 must trade off the reduction in sear
111. e operations on the displayed image logarithmic mapping of the magnitude It is displayed to better visualize the transform If you want to display the phase of your FFT image some direct calls toMIL are required You can use the Inspector script functions ImgGeRealMILID and mgGel magMILID to access thereal and imaginary MIL buffers that Inspector uses respectively The sample script phase bas illustrates how to display the phase of your image The techniques used in the phase bas script can also be used to implement complex arithmetic operations 136 Chapter 6 Advanced image processing Polar to rectangular and rectangular to polar transform Polar to rectangular and rectangular to polar transforms allow conversion of polar coordinates to cartesian coordinates and vice versa With Inspector you can perform rectangular to polar or polar to rectangular transforms using the Image Advanced Geometry Polar command Thefollowing is an example of a rectangular to polar transform The dotted lines define the borders of the zone of interest End radius Destination image Start radius End angle Start angle d Polar to rectangular and rectangular to polar transform 137 The result will be mapped to the destination image as shown below CBRE LK S RAT For arectangular to polar transform the borders of the zone of interest are defined by specifying the center the start and end radius and the st
112. e Number of Blobs field The results Number of Blobs will be assigned to the variable All of the robot s coordinates and blob features that were defined by the user will be put into the Results Table SP Set Position ViewFlex v 2 6 Sets coordinate value from the Results Table into the SCORBASE Position Number Set Position Ex T Cancel In the Position field insert SCORBASE Position Number In the Row field insert the Position Number Row in the Results Table In the Z mm field insert the Z coordinate by default the value in the Z coordinate will be the same as the value of the calibrated position Z can be a variable or constant 37 User Manual 0201 EF External Function Executes the function or subroutine that the user writes in the Matrox script External Function x r Script Name Function 4 r Parameter The results of the function will be assigned to the variable In the Script Name field insert the name of the script in the script folder In the Function field insert the function or subroutine name that is in the script In the Parameter field insert the parameter that can be the SCORBASE variable or constant Note Available for ViewFlex for SCORBASE for ER 4u only In the Return field insert SCORBASE variable that will get the function return GV Get Value ViewFlex v 2 6 0201 Receive a value from any cell in the Results Table
113. e display mapping mode see Chapter 3 Dealing with images Double buffering Grabs sothat thedigitizer grabs a frame into non paged DMA Host memory or on board memory when applicable while the previous frame is transferred to the display This is most useful if you want to display a continuous grab with a pseudo color palette for example the Blue Gray Red palette since images can only be associated with a palette if Windows is handling their display When 236 Chapter 13 Grabbing images and specifying camera settings performing a continuous grabin normal modeand displaying froman imaging board with a display section Windows is not handling the display Averaged Grabs and displays a running average on a basis of 8 frames where 7 8 of a given frame is accumulated data and the remaining 1 8 is data from the new frame During a continuous grab the camera grabs intoa livewindow and ends when you call the Grab Halt command If your system is not using an imaging board with a display section a continuous grab is performed in two stages During the grab the incoming data is transferred to the display in the fastest way possible given your system configuration When you call the Grab Halt command a final snapshot is madeinto system memory and this data can be used for further processing and display Due to these two stages the image quality might be different between the live grab and the final image Seethe on line pro
114. e expected by chance In our case wearenot interested in negative values so results are clipped to 0 In addition we use r instead of r to avoid the slow square root operation Finally the result is converted toa percentage where 100 represents a perfect match So the match score returned is actually Score max r 0 x 100 Note some of the terms in the normalized correlation function depend only on the model and hence can be evaluated once and for all when the model is defined The only terms that need to be calculated during the search are yt yr y This amounts to two multiplications and three additions for each model pixel On atypical PC the multiplications alone account for most of the computation time A typical application might need tofind a 128x128 pixel model in a 512x512 pixel image n such a case the total number of multiplications needed for an exhaustive search is 2x512x128 or over 8 billion On a typical PC this would take a few minutes much morethan the 5 msec or sothe 176 Chapter 9 Pattern matching search actually takes Clearly Inspector does much more than simply evaluate the correlation function at every pixel in the search region and return the location of the peak scores Hierarchical Search A reliable method of reducing the number of computations is to perform a so called hierarchical search Basically a series of smaller lower resolution versions of both the image and th
115. e image is identified by its position in the grid and is referenced by its row and column number Once images are represented digitally computers can reliably automate the extraction of useful information using digital image processing Digital image processing performs various types of image enhancements distortion corrections and measurements Image enhancement and transformation Inspector and image processing 35 Inspector and image processing Inspector provides a comprehensive set of image processing operations There are two main types of processing operations m Those that enhance or transform an image m Thosethat analyze an image that is generate a numeric or graphic report that relates specific image information nspector s image enhancement and transformation operations allow you toimprovethe quality of your image and or transform your image These include um Point to point operations These operations include brightness and contrast adjustment constant thresholding arithmetic and image mapping operations Point to point operations computeeach pixel result as a function of the pixel valueat a corresponding location in either oneor twoimages a Spatial filtering operations These operations are also known as convolution They include operations that can enhance and smooth images accentuate image edges and remove noise from an image M ost of these operations compute results based on an underlying ne
116. e image types such as 16 and 32 bits including 32 bit float Such images are useful for storing intermediate processing results since they can contain a wide range of values You can also use 16 bit images tostore data grabbed from 10 or 12 bit deep camera sources Notethat all non 8 bit single band images can only be saved as mim files If you want to view a histogram of a deep image that has only a small range of values you can select the View Zoom Interactive command to draw the histogram with the small subset of the dynamic range of values in your image See Performing a histogram of Chapter 8 Histograms profiles and measurements Packed binary images Inspector also supports packed binary images Generally processing operations are optimized for packed binary images especially morphological operations With all processing operations you can specify if you want the results in a packed binary format Packed binary images can only be saved in mim format Displaying grayscale images Regardless of your display resolution image depths other than 8 bit unsigned or 3 band 8 bit color will require remapping to the display I nspector displays these images by mapping them to a separate display buffer When creating a new image with a type and depth other than 8 bit unsigned you will be presented with a dialog box where you can select a mapping method 82 Chapter 3 Dealing with images m Autoscale Theautoscale
117. e lowest highest value respectively of the output range By stretching the range of pixel values window leveling allows you to make use of all available intensities Example 3 Working with many images and non standard image file types 65 1 Toselect the appropriate window leveling settings for the entire vascular bed not just a small ROI expand the small ROI sothat it encompasses the entire vasculature for all frames in the sequence To dothis select the ROI R SAMPLE DCM OJ anddragoneof its handlestomake a rectangle of the required size 2 Now click on the Window Leveling button to remap the range of pixel values in the larger ROI Alternatively use me the Image Processing Mapping Window Leveling command As a third alternative right click on the active sequence and choose Processing from the context menu The Processing Operations dialog box appears Select the Mapping tab of the Processing Operations dialog box The WL Window Leveling button will be selected by default Live preview and Determining the output pixel value range that produces the locking the live preview best contrast in your image might be difficult However Inspector allows you to obtain a live preview of any mapping modifications and maintain this live preview over all the required frames in a sequence 66 Chapter 2 A few tutorials 1 2 3 a v 4 m 5 Ensurethat theOutput Rangeis set for thefull pixel value
118. e operation set the center pixel to 1 otherwise leave as is neighborhood match the corresponding kernel value Operation Binary Grayscale ERODE If the kernel does not Subtract each kernel value from the match the neighborhood __ corresponding pixel value in the exactly set the center pixel jneighborhood then replace the center to 0 otherwise leave as is pixel of the neighborhood with the minimum value from the resulting neighborhood values DILATE If any of the elements of the Add each kernel value to the corresponding pixel valuein the neighborhood then replace the center pixel of the neighborhood with the maximum value from the resulting neighborhood value Creating custom structuring elements 129 THIN If a pixel s neighborhood If MAX O center pixel lt MIN 1 matches the kernel exactly set center pixel to MAX 0 otherwise set it to 0 otherwise leave leave center pixel as is MAX 0 is the as is maximum of all pixels in the neighborhood that correspond to 0 in the kernel and MIN 1 is the minimum of all pixels in the neighborhood that correspond to 1 in the kernel THICK If a pixel s neighborhood If MAX 0 center pixel lt MIN 1 matches the kernel exactly iset center pixel to MIN 1 otherwise set it to 1 otherwise leave leave center pixel as is MAX 0 is the as is maximum of all pixels in the neighborhood that correspond to 0 in the kernel and
119. e that the picture is adjusted clear and free of distortions Click the Calibration icon This automatically establishes a connection between controller and robot Move robot arm into desired position within the camera s field of vision Using a pen or marker mark a position on a piece of paper or on the coordinate grid close to the robot and strong enough to be seen in the 27 User Manual ViewFlex v 2 8 0201 camera picture This position should be aligned with the TCP tool center position the exact center of the robot gripper This first position is called Pos 1 Calibration Step 1 dialog box Click Record Position 1 f using Controller B be sure it is in Auto Mode If trying to record a position produces a beep sound the reason may be that the Teach Pendant has not been changed from Teach Mode to Auto Mode In the ACL Terminal type Auto then click Enter Only along the X axis move the robot to a position on the X axis that would be greater X than Pos 1 Take a writing instrument and mark the position This second position is called Pos 2 Click Record Position 2 f using Controller B be sure it is in Auto Mode Now you will see in the Combo box the distance between the two positions Setting Distance on the X axis To achieve the most accurate measurement take a ruler and measure the distance between Pos 1 and Pos 2 by hand Change the distance if necessary according to the h
120. each pixel with the minimum value in its 3x3 neighborhood if you areoperating on bright objects or with the maximum value if you are operating on dark objects When using binary mode Inspector s predefined erosion operation uses an AND operation with thestructuring element When using grayscale mode I nspector s predefined dilation operation replaces each pixel with the maximum value in its 3x3neighborhood if you areoperating on bright objects or with theminimum value if you are operating on dark objects When using binary mode Inspector s predefined dilation operation uses an OR operation with the strucutring element Distance Thick Thin Other morphological operations 123 If the predefined erosion and dilation operations do not suit your needs you can create your own structuring el ement for either of these operations Other morphological operations You can also perform distance thick or thin operations using the Image Processing Morphology command Note that these morphological operations only operate on bright objects If your objects are dark you can invert your image for example through a thresholding operation before performing one of these operations Inadistance operation each pixel s valueis determined by the radial distance from that pixel to the background the greater the distance the brighter the value assigned tothe pixel Thresholding the result of a distance operation can be used to remove
121. ecified in the registry is used to run Inspector To disable the System Allocation list box usethe Options Preferences command in the General tab System Allocation METEOR 2 30 Chapter 1 Getting started Inspector doc umentation conventions This section describes some conventions used in this manual Menu conventions This manual referstooperations accessiblethrough the menus as commands and refers to them by their menu and sub menu names F or example selecting Geometry fromtheTools menu and then selecting Scale from the presented sub menu is referred to as the Tools Geometry Scale command Dialog box conventions In general when a command is called a dialog box appears Dialog boxes allow you to specify how you want the command performed To use a dialog box you might have to enter values in edit fields select items from list boxes specify options or dick on buttons and selectors Calibration dialog box Apply to Apply to options histla tif Units drop down list box Units pixels ecc Cancel Scaling Square pixels checkbox v Square pixels Pixels World PA l L Default Pixels edit field Origin Angle a s Oe Angle button Y Jo Angle An ordinary list box differs from a drop down list box in that the list is immediately visible To view the contents of a drop down list box dick on the arrow next to the box Inspect
122. ed in the Util directory Chapter 7 Analyzing images This chapter describes how you can analyze your images using Inspector 140 Chapter 7 Analyzing images Analyzing images with Inspector Once you have improved and transformed your image using previously discussed techniques you can begin toanalyze your image In other words you can begin to extract useful information from your image This extracted information might be required to perform some subsequent operation in nspector or in another software package Alternatively the extracted information might be required to summarize the effect of an operation Theinformation to extract can be basic such asthedistribution of pixel values in an image histograms a line profile X and Y profiles the angle between specified markers the convex perimeter of blobs or the location of a model within an image The information can also be of a more advanced nature such as the population distribution of blobs based on specific features the dassification of blobs in pre specified types blob analysis trends and statistics over time using several measurements and finding and measuring edges Each of these operations is described in detail in subsequent chapters Aspects common to these operations such as the Result Log window and the measurement tableare discussed in this chapter Tomakesummarizing your results easier you can export them save them to a file print them
123. efour methods of creating a new sequence Creating a sequence using the File Grab Sequence command was discussed in the previous section Grabbing sequences There are three other methods of creating sequences using the File New command m You can insert image files into a new sequence using the Sequencel nsert F rame command nserted images areadded tothe end of the sequence When inserting multiple image files at the same ti me they are added to the end of the sequence in reverse order that is the last file selected is inserted first and the first one selected is inserted last Sequence frames once inserted can easily be rearranged by dragging and dropping the thumbnails See the sample script Buil dsequence bas as an example of creating a sequence and sorting frames by some criteria You can drag and drop open image windows intoa new empty sequence You can drag and drop frames from existing sequences that are open into a new empty sequence Note that the images must be the same type size and depth If you try toinsert an image whose attributes do not match those of the frames already in the sequence an error message will appear If you drag and drop frames from an open memory sequence the frames are moved to the new sequence not copied If you want to copy the frames usethe Ctrl key and check for the small plus sign on the mouse cursor when you drag to the new sequence window You can drag and drop multiple fra
124. eful toenhance edges in a certain direction and removethose in another Laplacian edge detection f The Laplacian operations emphasize the maximum values or peaks within an image Point to point operations Arithmetic and logical operations Arithmetic with images 121 Arithmetic with images Itis often very useful to performarithmeticor logical operations on images These operations apply the specified operator on individual pixel values in a source image or on pixels at corresponding locations in two sourceimages When the source images are color the operation is applied on each component Operations whoseresults do not depend on neighboring values are known as point to point operations Arithmetic and logical operations fit into this category Using the Image Processing Arithmetic command you can perform the following arithmetic and logical operations m Add subtract multiply divide AND NAND OR XOR NOR or XNOR twoimages or an image and a constant You can also add and subtract with saturation Takethe absolute difference between two images m NOT negate or take the absolute value of an image Scale and offset an image Perform a bit shift left or bit shift right operation Fill an image with a specified value m Copy the contents of an image to another image Note that iftheimages are of different types the data is not converted or scaled To copy with conversion or scalin
125. el can have up to seven rows and seven columns Kernels can include don t care values that is values that you want the operation to ignore 128 Chapter 6 Advanced image processing When creating the kernels of the structuring element several predefined patterns are available These indude circle cross and prune as well as patterns based on the Golay Alphabet If the kernel is toolargefor the selected pattern the pattern will only appear in theupper left corner of the kernel If the kernel is too small for the selected pattern the pattern won t be fully represented You can create a structuring element that includes rotated versions of a kernel n other words you can create one kernel then have Inspector automatically rotate it by 90 180 and 270 intokernels 2 3 and A You can alsohavelnspector rotate two kernels in which case kernel 1is rotated by 90 180 and 270 into kernels 3 5 and 7 kernel 2 is rotated by 90 180 and 270 intokernels 4 6 and 8 For moreinformation seethe on line procedure To create a structuring element Overview of morphological algorithms Thefollowingsummarizesthebinary and grayscalealgorithms used for I nspector s morphological operations Note that the hit or miss and match operations usethe same algorithm Also note that morphological operations are not recursive even if the source and destination images arethe same I n these cases an internal image is kept to perform th
126. eld of vi ew You can havethe positi ve Y axis oriented up 909 counter clockwise with respect to the positive X axis by default Inspector assumes it is oriented down 909 clockwise You can apply the calibration object to your current image which is your grid or to all new images If you want to apply thecalibration toan already opened i mage savethecalibration object and then load and apply it to the opened image Piecewise linear interpolation Perspective transformation Calibrating your imaging setup 259 Calibration modes When you calibrate from a grid or a list of coordinates see Calibrating using coordinates in an image you also haveto specify the calibration mode Inspector supports the following calibration modes m Piecewise linear interpolation m Perspective transformation n general you should use piecewise interpolation mode This mode can compensate for any kind of distortion It is very accurate for points located inside the working area the area of the image from which you want real world coordinates However it isless accuratefor points outsidethe working area The piecewise linear interpolation mode fits a piecewise linear interpolation function to the set of image coordinates and their real world equivalents The perspective transformation mode can compensate for rotation translation scale and perspective distortions For such distortions the perspective transformation mode is accu
127. en an edge and its background using the marker contrast setting Contrast is useful in distinguishing between several different edges particularly when the required edge does not have the largest edge strength described later or when the edge is at an angle The contrast for a stripe marker requires two values one for each of the stripe s edges Width of an edge marker Width of a stripe marker Markers advanced characteristics 215 Markers advanced characteristics Thissection describes the advanced characteristics of edge and stripe markers that are available in the Edge and Stripe Settings dialog box or returned as measurement results in the measurement table Width Edges are usually gradual shifts in grayscale values over several pixels The smoother the image the more gradual the change The width of an edge can be seen as a measure in pixels of this gradual shift in grayscale values The diagram below illustrates a profile of an edge where the gradual transition from black to white can be seen Edge position White gt Edge transition curve Black I Edge width nspector calculates the marker s position to be in the middle of this width Note the morean edgeis at an anglethe greater the stretching or distortion of its actual width Tohelpfindastripemarker you can specify thetypical distance between both of its edges in pixels Thedefault is 0 which finds a stripe with any width 216
128. epla mar ELG WALH Ee zm one nren e s Look at thepixel intensity values Thepill pixel valuesarequite distinct from the background pixel values Moreover the edge width and edgeangle between theforeground and background is well maintained Therefore the edges should be sharp enough sothat a singlethreshold value can be chosen that will properly segment your foreground objects from the background 46 Chapter 2 A few tutorials Undoing a processing After an enhancement or transformation operation has operation modified an ROI such as the smoothing operation just performed the operation can be undone by clicking on the i Undo button or using Inspector s Edit Undo command Note you can specify the number of operations that can be undone in the General tab of the Options Preferences command Notealsothat an operation that places results in a new window cannot be undone Toreverse the undo dick on the Redo button or usethe Edit Redo command Blob analysis Now perform a blob analysis operation on thegrayscaleimage so that you can determine if there are the correct number of pills in the blister package The first step will be segmentation that is providing a pixel intensity threshold value such that pill objects blobs are white foreground object and the background is black Then you will count these blobs as well as obtain some statistical information about each blob New command or alternatively select
129. er of collections that can be opened at any onetime depends on the amount of free memory in your system 96 Chapter 3 Dealing with images mage representations in a collection can be organized by dragging them from one position to another or by using the available sort commands n addition they can be dragged or copied between collections See the on line procedure Organizing image representations From the collection window you can also manage image files on disk You can open images by dragging them from the collection window onto the Inspector workspace or delete images Once an image is viewed in its full size you can manipulateit as you would any other I nspector image You can alsochangethe current path of image files in your collection by using the Collection Modify Paths command This is particularly useful when you areloading images from network drives Thecollection can be associated with a description presumably the image theme of the collection You can also associate a description with each image representation within a collection to distinguish the various images When copying an image representation its associated description if any is also copied You can then change the descri ption of the new image representation without affecting the description of the original image representation See the on line procedure Identifying and maintaining collections Zooming 97 Zooming With Inspector you can zoom
130. er physically resizing in that the operation takes less time to perform When you calibrate an image however you do not alter its size sothe image still appears distorted If the physical appearance of an image is important you might want to physically resize it Inspector provides some basic geometrical transform operations to improve your image such as scaling rotation translation and symmetry n addition for more complex distortions Inspector also allows you to transform your image based on its associated calibration Calibration is discussed in Chapter 15 Calibration Scaling Scaling allows you to correct for aspect ratio distortions in an image or reduce or magnify the image to the appropriate size To scale your image you can use the Image Processing Geometry Scaling command You can resizethe image along the horizontal and or the vertical axis Rotating images In some instances the orientation of an image can lead to erroneous conclusions When an image has been rotated from its original position you can realign it using the Image Geometry Rotation command Center of rotation Basic geometrical transforms 113 When rotating an image Inspector allows you to specify the center of the rotation The center of rotation is the position within the image around which the rotation takes place The default center of rotation is the center of the image The diagrams below display the effect of rotating an image by 6
131. ershed transformations Chapter 7 Analyzing images Analyzing images with Inspector Result log window esses Displaying results asr ane as The measurement table Calibrator e ere a e a a eld RUE RUE bes Exporting results leen Exporting results 0 0 Exporting graphics Keeping analysis statistics Pixel conventions lle Chapter 8 Histograms profiles and measurements 147 Performing a histogram 148 Histogram window ss 148 Examining pixel values using profiles 149 Profiles of line and other graphic objects llle 149 Outline thickness 0 150 X and Y profiles of aregion 150 Profile window esee 151 Adjusting profiles 151 Features of histogram profile graph windows 0 0 00 008 152 Taking measurements using graphic objects 0 0 00 00 008 154 Finding average intensities using graphic objects 154 Taking measurements between result objects 0 0 0 0 0 000008 155 Chapter 9 Pattern matching 5 157 Pattern matching 0 158 Steps to performing a search 159 Defining a search model 160 Saving a model sess 160 Fine tuning the model
132. erveCam on another computer with USB connection that is also connected to a network 16 A driver message will appear Click OK Software Initialization The ViewFlex ini file initializes the ViewFlex software Eshed RobotecSViewFlexNViewFlex exe ViewFlex ini C Program FilesXE shed RobotecXViewFlex ViewFlex Properties The ViewFlex ini file includes the following The directory in which the script files are located only for the ViewFlex Device Driver Device Driver Definitions ScriptPath Vision Script You can choose the controller type A B PC USB ViewFlex v 2 8 11 User Manual 0201 Controller Setting ControllerzA Note You can choose ER4 ER4u ERS ER9 ER14 SV RobotType ER14 Activating ViewFlex To start the ViewFlex software double click the ViewFlex icon on the desktop View Flex ViewFlex OR From the Start menu select Programs ViewFlex ViewFlex Activating ViewFlex for SCORBASE for ER 4pc To start ViewFlex for SCORBASE do the following 1 Activate ServeCam 2 Double click the SCORBASE icon Or double click the ViewFlex for SCORBASE icon amp C WINDOWS Start Menu Programs ViewFlex for Scorbase Jol xf Ele Edit View Go Favortes Help le a SY 23 3 x sl Address fien C AWINDOWS Start Menu Programs ViewFlex for Scorbase e mw SCORBASE f Setup PC Uninstall ViewFlex for for C ll ViewFlex f Scorbase SCORBOT ER CORO 4 object
133. es Target image Model Rotated model Model rotated using circular overscan Masked pixels Inspector s implementation of a circular overscan search is faster and more accurate than that of a rotated model search when performing an angular search A rotated model search Rotation 169 should only be used when the pixels surrounding the ROI follow no predictable pattern for example when searching for a bolt in an image of loose nuts and bolts Rotated model search To perform a rotated model search follow the same procedure as you would when manually creating the model using the Analysis Pattern New Model command making surethe Use Circular Overscan command is disabled Check the Enable search with rotate checkbox found in the Angletab of the Model dialog box to set the search angle of the model and the Delta negative and Delta positive of the angular range if necessary discussed in greater detail later in this section Angles are expressed in degrees and floating point values are accepted When a non circular overscan model is preprocessed nspector will internally create different models by rotating the original model at therequired angles masking the corners of the model automatically so that these pixels are ignored during the search Note you can also use the New Rotated Model command to search for a model at one specific angle However in this case you should not also specify a range of angles
134. es to improve images 105 Techniques to improve images If interference problems cannot be removed at the source preprocessing might berequired toimprovethe image as much as possible without affecting the information that you are seeking There are several techniques that you can use to improve your image Adjusttheintensity distribution of your image by performing a window leveling or a histogram equalization operation or by changing the image contrast and brightness Reduce Gaussian random noise and systematic noise with small scale variations by applying a low pass spatial filter to your image This technique replaces each pixel with a weighted sum of its neighborhood Alternatively to reduce Gaussian noise you can perform a frame averaging type grab Remove salt and pepper noise by applying a median filter to your image This technique replaces each pixel with the median pixel value of its neighborhood Perform a morphological opening operation to remove small particles and break isthmuses between objects in your image Perform a morphological closing operation to remove small holes in objects in your image Remove systematic noise by determining the pattern s frequencies of the noise and filtering out these patterns This can be done using an FFT transform Reduce object shape distortions by making surethat thetype of camera you allocate digitizes the image with squarepixels that is a 1 1 aspect ratio If this is no
135. esults can be read and acted upon immediately If a minimum and maximum value are specified results are compared against these values and graded a pass or fail These pass fail tests are summarized in the Result Log window If the expected value is specified and the Ver bose logged results option is enabled the mean variation tothenominal is also returned in the Result Log window Note that if a result does not pass it is still included in thestatistical results Todetermine appropriatetolerance values you should analyze a few typical images and look at the resulting mean and standard deviation The mean will give you an idea of what 146 Chapter 7 Analyzing images An example expected value to use the deviation from this mean will give you an idea of what minimum and maximum values to use In general the more sample images you have the more appropriate your specified tolerances Note that tolerances can be set directly from the graph of the appropriate measurement statistic object or blob feature Along a pharmaceutical inspection line you might want to assess the quality of each package by calculating the diameter of each pill in the package If a package has a pill with a diameter below a certain value the package is not suitable for shipment If you specify the acceptable range of values for the diameter nspector assesses each pill against these values and grades a pass or fail Pixel conventions The center o
136. et Y You can also determine the minimum maximum and mean F eret diameters of the blob Feret Min Diam Feret Max Diam and Feret Mean Diam Thelength of a blob can then be defined as its maximum Feret diameter and the width of a blob as its minimum F eret diameter Diameters are determined by checking the F eret diameter of the blob at a specified number of angles Increasingthenumber of angles tested increases the accuracy of the results but also increases the amount of processing time Note that the maximum F eret diameter is not very sensitive to the number of angles and 8 angles usually gives an accurate result With Inspector you can also determine the angles at which the minimum andthemaximum F eret diameter werefound Feret Min Angle and Feret Max Angle and the ratio of the maximum to minimum F eret diameter Feret Elongation Notethat the minimum and maximum F eret diameter of along thin blobis not very representative of its dimensions especially if the blob is curved This is true even if the minimum and maximum F eret diameter are checked using the maximum number of angles 64 For long thin blobs andlong thin curved blobs the following features are better a Length m Breadth These features are derived from the area and perimeter assuming that the blob area is equal to the length x breadth and the perimeter is equal to 2 length breadth Note that these calculations are not valid for most blob types ho
137. exactly the same as a convolution i N r X m i l In other words for each result the N pixels of the model are multiplied by the N underlying image pixels and these products are summed Note the model does not have to be rectangular because it can contain masked pixels that are completely ignored during the calculation When the correlation function is evaluated at every pixel in the target image the locations wheretheresult is largest are those where thesurroundingimageis most similar tothe model The search algorithm then has to locate these peaks in the correlation result and return their positions Unfortunately with ordinary correlation the result increases if the image gets brighter In fact the function reaches a maximum when the image is uniformly white even though at this point it no longer looks like the model The solution is to The pattern matching algorithm for advanced users 175 use a more complex normalized version of the correlation function the subscripts have been removed for clarity but the summation is still over the N model pixels that are not don t cares Yo 2 2 2 2 xo bo With this expression theresult is unaffected by linear changes constant gain and offset in the image or model pixel values The result reaches its maximum value of 1 where the image and model match exactly gives 0 where the model and image are uncorrelated and is negative where the similarity is less than might b
138. f a pixel is important for all Inspector functions which return positional results with subpixel accuracy The reference position of a pixel is its center and the resulting subpixel coordinates are with respect to the pixel s center With thisin mind the coordinates of the center of an imagecan always be found using the following formula Width 1 Height E 2 2 For example the following image contains 4 pixels If the formula is applied the center of the image is found at 1 5 0 Chapter 8 Histograms profiles and measurements This chapter describes different basic techniques for examining pixel intensities in an image 148 Chapter 8 Histograms profiles and measurements Histogram graph Performing a histogram A histogram of an image is the pixel count of each color component intensity in theimage You can obtain thehistogram of an image using the Image Statistics Histogram command There are three main reasons to perform a histogram m Toexaminecontrast in an image This allows you toseeif the full intensity rangeis being used especially when operating on a grayscale image m To choose thresholding values to segment objects from the background Notethat if you only need onethreshold value you can choose to autothreshold instead of performing a histogram and then the threshold operation m To determine the type and amount of noise present in what should be a flat single colo
139. g usethe Image Convert command For more information see Performing a color or data type conversions in Chapter 3 122 Chapter 5 Image transformations Erosion Dilation Erosion and dilation Erosion and dilation are neighborhood operations that determine each pixel s value according to its geometric relationship with neighborhood pixels and as such are part of a group of operations known as morphological operations m An erosion peels layers from objects or particles removing extraneous pixels and small particles from the image a A dilation adds layers to objects or particles enlarging particles Dilation can return eroded particles to their original size but not necessarily to their exact original shape To perform an erosion or dilation operation use the Image Processing Morphology command Erosion and dilation can beused tofindtheperimeter of objects If you XOR the eroded or dilated version of an image from the original image you obtain the endoskeleton or exoskeleton of the object s perimeter Note that erosion and dilation are the basic operations used to perform the opening and closing operations discussed in the previous chapter You can accelerate processing for erosion and dilation operations operations by selecting the Binary mode or No Escape options which is discussed in the Opening and closing section of Chapter 4 When using grayscale mode Inspector s predefined erosion operation replaces
140. g a grab to disk is supported click the Compression button when specifying the disk sequence s filename however the grab will be slower If you have specialized hardware it will be used Unless you have a fast machine with an optimized CoDec a grab to disk is most suitable for time lapse applications with a large number of frames being collected at a relatively low rate for example growth studies waiting for mechanical displacements or surveillance Processing the frames of a sequence 251 Processing the frames of a sequence Frames of a sequence can be used as a sourcefor any processing or analysis operation If your sequence is a memory sequence you can annotate process into or rearrangethe frames in the sequence Usethe sequence playback tools to select the frame or select the frame directly and perform the required operation You can also activate the frame by dicking on it or using the scripting function SegSetCurrentF rama m If you have a disk sequence which you must analyze you can create a new memory sequence using the File New command and paste the frames from the disk sequence into thenew memory sequence Then theframes can be processed like any other image Frames of any sequence can be copied to the dipboard or duplicated like any other image and you can create images from frames of any sequence type To create an image from a particular frame in a sequence double click on it to bring it up as an
141. g features you will probably want to analyze results All blobfeatureresults aredisplayed in theAll Results table To view the results for a single blob double click on the blob within thesourceimage Results for that blob arereturned in the One Result table You can also analyze results using the statistics of the blob analysis operation You can view the statistics for all blobs or only for those blobs of a specific class For more on statistics refer totheStatisticsand tolerancessectionlater in this chapter Onceyou havetherequired results you can export theseresults and or transfer positional results to the measurement table Totransfer positional results to the measurement table select therow s totransfer andthen click on the Transfer button If the blob result table does not contain any positional results the Transfer button is disabled Saving and loading blob settings Once you have determined the appropriate features for calculation and selected your segmentation mask setting you might want to save these settings After saving you can load the blob settings from disk toanalyze other images In this case you won t need to add or remove features or re define blobfilters or classes n addition you can apply these blob settings to several typical imagestodetermine appropriatetolerances for automated checking When you load thisfileatalater time theBL OBSET dialog box appears with each tab exactly as it was when you
142. ge None When loadingor creating an image you can also select not to display the actual image data the image window displays a cross hatch instead of image data This method is useful if the image is to be used for processing only Note that non displayable images do not support graphic annotations ROIs and line profiles Note that even if a display buffer is used to display the image the displayed image s status bar shows the actual value of the pixel under thecursor If theimageis associated with a palette the mapped RGB component values or gray values if using a grayscale palette will also be displayed in the image s status bar Grayscale or single band images 83 You can changethe display method at any time using the View Properties command Displaying single band images with a palette Inspector displays any single band image with a display palette in order to best represent the ideal colors of the image Palettes map an image s pixel values to required display values the image s data is not directly affected Palettes can adjust the color and even improve the image for visualization purposes this is sometimes called pseudo coloring By default the palette associated with an image is linear gray You can view a 1 band image a component of a color image or a panein 3D view in another color scheme by associatingit with an Inspector palette Although palettes are not available for color images viewed in composite
143. ge you should first convert it to RGB to produce meaningful results If your system cannot display quantized images you can use the system bar and drag the image to the VGA system Colorimages Color images include 8 bit 3 band images such as RGB and HSL true color images Multiple views Inspector provides a very flexible 3 band color image display Seethe View menu Color images can be displayed in m One window as a composite image using the View View Composite command m Threeseparate windows where each window displays one of the image s color components using the View View Components MDI command One window divided into three panes where each pane displays one of theimage s color components usingthe View View Components 3 command Onewindow divided intofour panes whereonepanedisplays the composite image and the remaining panes display each of the individual color components of the image using the View View Components 4 command Even though the latter two view modes are part of the same window each pane once selected can be resized and have its contents zoomed and scrolled independently Select a pane by clicking on it the active pane is identified in the image s title bar Note that the palette of the active pane is favored when displaying the image with its own palette If a window or pane displays a single component you can change its display palette using the Palette Editor This command is dis
144. hat kernel 2 It then combines the results of each kernel Inspector can either sum the results or take the maximum value of the results 3 Thefinal offset divide and absolute value options defined for the filter are then applied 4 If enabled results are saturated If not enabled overflows or underflows will wrap around For more information on how to filter your image see the on line procedure To createa filter Creating custom structuring elements A morphological operation is a type of neighborhood operation that determines the new value for a pixel based on the relationship between its neighborhood and a given set of numbers The numbers are often represented in a matrix Known as a structuring denent The type of morphological operation and the structuring element determine the effect of the operation With Inspector you can create your own structuring element based on one of eight types of morphological operations erode dilate thin thick hit or miss match open and close using the Image Processing Morphology command and clicking on the Edit button The algorithms used by these operations are summarized in the Overview of morphological algorithms section A structuring element can contain up to sixteen kernels that is sixteen matrices except for the open and close operations which use only one kernel In binary mode a kernel can have up to sixteen rows and sixteen columns In grayscale mode a kern
145. he ViewFlex Allows you to test and debug the Vision Machine process Emulates the Vision Machine System in Simulation mode 30 User Manual ViewFlex Properties 2 x General Shortcut Netware Version ViewFlex E rix Target type Application Target location ViewFlex Target JEshed Robotec WiewFlex ViewFlex exe ViewFlex ini Start in C Program Files Eshed Robotec ViewFlex Shortcut key None Bun Normal window bi Eind Target Change Icon Close Gercel Apply ViewFlex Properties ViewFlex exe After the ViewFlex is installed copy ViewFlex exe to the OpenCIM Bin folder ViewFlex ini Be sure to update the script path to the folder where the scripts files are An example of where the directory that the script files are located in Device Driver Definitions ScriptPath C opencim32 lib ViewFlex Operation Modes The Operation Mode lets you to define the control mode ViewFlex Device Driver is running in On Line On Line mode the ViewFlex Device Driver waits for the commands from the OpenCIM Manager and then Snap Load Script Execute Program Script and Send Results Manual Manual Mode allows you to test the OpenCIM Manager and manually simulate conditions of Fail and Pass The following are activated only in Manual Mode after a request from the OpenCIM Manager y Send Pass Sends pass to the Manager ViewFlex v 2 8 31 User Manual 0201 x Send
146. he grab affects whether successive frames can be captured a high speed grab can grab more successi ve frames than a grab at a lower speed Grabbing to paged Host memory When you grab to paged H ost memory system RAM thespeed of the grab depends on the transfer rate from your frame grabber to computer memory This in turn depends on the following m The maximum transfer rate from the frame grabber to the PCI bus T hePCI controller on the path to computer memory that is Neptune Triton or Orion chip The amount of memory available for the grab depends on the combination of the amount of real memory and the size of the swap file Grabbing to on board or DMA non paged memory Grabbing to on board memory or DMA memory non paged H ost memory is generally faster than grabbing to paged H ost memory or disk As a result you areless likely to miss frames However on board memory for example on Matrox Genesis islimited Theamount of DM A memory istheamount specified when you installed Inspector If you do not have sufficient free memory for the requested number of frames you can choose to store some of your images in another memory resource using the System toolbar F or more information on the System toolbar see the on line help topic The System Bar 248 Chapter 14 Creating and manipulating sequences Once you have grabbed a memory sequence you can save it to disk but you must ensure that all modificat
147. hic representations of analysis operations copy them from an active window in Inspector to the clipboard and paste in the other application Exporting graphical representations of data might be useful when documenting results Statistics graph Tolerances Keeping analysis statistics 145 Keeping analysis statistic s When performing blob analysis or measurement operations Inspector maintains statistics on results For example if you are calculating the area of blobs in an image Inspector maintains such statistics as the mean and minimum area For blob operations statistics are automatically maintained For measurement operations only those which you specify are maintained Each time results are calculated statistics are updated You can view and or reset these statistics You can view statistics in a trend or distribution graph A trend graph displaysthelast n featureor measurement results that were kept A distribution graph displays the number of results kept that fall into each range of result values Notethat a trend graph allows you to for example notice a failure before it becomes critical A distribution graph allows you to determine dassification ranges for different objects in an image You can compare results against specified tolerances Specifically you can compare results against m A minimum and maximum value m An expected value Setting tolerances can be useful during automation since r
148. hoose the required keypoint Calculation results are automatically placed in the measurement table as new result objects 156 Chapter 8 Histograms profiles and measurements Chapter 9 Pattern matching This chapter describes how to define a pattern matching model and how to search for that model in an image 158 Chapter 9 Pattern matching Pattem matching In many applications it is useful to know where in an image something is located In a machine guidance application for example a mechanical device might need to know where mounting holes arelocated on a circuit board sothat screws can be inserted properly In the semi conductor industry finding fiducial markers is required to determine the orientation of objects n these cases exact positions must be calculated a minor deviation could cause problems during assembly With I nspector s pattern matching operations you can perform such searches quickly reliably and with sub pixel accuracy These operations areimplemented using normalized grayscale correlation nspector refers to the pattern for which you are searching as the search modd and the image from which it is extracted as the modd s sourceimage The image being searched is called the target image J Model Model s source Image Toobtain the best search model you should test several models created from images acquired over the full range of expected conditions for the target image
149. icon 25 User Manual 15 x Pattern Matching Model dialog box From the Dimension tab you will see the cube that was outlined using the ROI icon From the Search tab click All From the Angle tab check Enable Search With Rotate Then enter 180 for Delta Negative and 180 for Delta Positive this enables a full 360 degree search a Click Search This will build a Measurement Table containing all the positions of the object relative to Pos 1 PattemPos 1 1744 12275 nomzod Pattem Pos 2 11778 21457 95595 an00de Pattem Pos 3 23775 2753 9418x 50004 Patem Pos 4 23830 045 9373x 25008 Delculate Piete heise Measurement Table ViewFlex v 2 8 26 User Manual 0201 Calibration ViewFlex v 2 6 0201 3 IMAGE2 8U 351 287 laray 91 7 Searched Images In this example choose the Current table which will be the opened table and Position 1 Pos 1 Click OK The robot will move according to the Go To Position and find the cube T Synchronizing the Vision System and the Robot To calibrate real world measurements with the Vision System s world of pixels it is necessary to perform the following steps l Make sure that there is communication between the controller and the robot Click the ACL Terminal icon Click the On Line Off Line icon if necessary Execute the HOME routine Open Camera 1 Check camera to make sur
150. iders left and right to decrease and increase the tones of the colors in the video This option is not available when the video source is an external video device Click Restore Defaults to set the sliders to the factory settings for this tab White Balance White balance is not available when the video source is an external video device 17 User Manual Advanced ViewFlex v 2 6 0201 Automatic White Balance Automatically makes the whitest color in the video whiter and adjusts the other colors proportionately Move the slider to the left to make the video appear cooler adds blue move the slider to the right to make the video appear warmer adds red Any changes made take a few seconds to happen Manual White Balance When switching from automatic to manual the settings from the automatic white balance control are initially matched until you change them Move the slider to the left to make the video appear cooler adds blue move the slider to the right to make the video appear warmer adds red Manually adjusting the white balance may be useful when using a laptop computer in differing lighting conditions for instance going from room lighting to sunlight or when automatic adjusting is unsatisfactory Image Effects Click Mirror Image to reverse the image in the video window Click Flip Image to have the image flip from top to bottom This could be used when the camera is upside down for instance when mounted underne
151. if the image is zoomed out If you want to view an image at its original size hold the Ctrl key and click the left mouse button to shrink to the original size or dick the right mouse button to grow to the original size Toreturn to normal viewing mode press on E sc or click on the Zoom button again 98 Chapter 3 Dealing with images Frame size When zooming in the image s frame is normally expanded as theimageis zoomed up to a maxi mum size If you do not want theimage s frameto expand as you arezoomingin on theimage select the View Properties Display tab and deselect the Frame Expands option from the tab Accessing context menus in zoom mode Note that in zoom mode pressing the right mouse button will normally zoom the image out if possible To access the image s context menu in zoom mode hold the Shift key and dick the right mouse button Special scrolling mode In addition to using the scroll bars to view different parts of an image you can press the Shift key while dragging the cursor over theimage This mode allows you to movethe entire image diagonally rather than scrolling in one direction at a time View as text With Inspector you can display the pixel values of the current imageas a series of hexidecimal numbers in an adjoining pane usingthe View View as text command This featureis useful if you need to determine the numeric value for a given color at any point in the image you can also precisel
152. ighborhood process the weighted sum of a pixel value and its neighbors values You can use one of Inspector s pre defined filters or create your own filter s Morphological operations These operations include erosion dilation opening and closing of images These operations compute new values according to geometric relationships and matches to known patterns structuring elements in the source image If necessary you can create your own structuring element for one of I nspector s mor phologi cal operations Geometric operations These operations include scaling rotating translating and flipping of images as well as polar to rectangular and distortion correction 36 Chapter 2 A few tutorials Image analysis nspector s image analysis operations summarize a frame of pixels into a set of values which relate specific image information These include Basic statistical operations These operations extract statistical information from a specified region of an image such as its distribution of pixel values or its profile Pattern matching operations These operations allow you to determine how similar certain areas of the image are toa pattern Blob analysis operations These operations allow you to identify and measure aggregated regions of pixels within an image commonly known as blobs Measurement operations These operations allow you to locate edges or pairs of edges known as stripes and measure their fe
153. image Note that when you create an image from a sequence frame its identifier changestol IMAGE TheEdit Copy command or the Duplicate command will alsocreate an image The image is independent of the sequence that is changes in the image do not affect the frame in the sequence Notethat if aframeof a sequenceis used as a destinati on image the sequence must be a memory sequence if you want to save the results of the operation sequences grabbed to disk or opened from disk cannot be modified 252 Chapter 14 Creating and manipulating sequences Inspector supports dynamic update of histogram and line profile graphs when playing back the frames in the sequence When you want mapping or thresholding preview to apply to all theframes in the sequence select the Lock Live Preview button of the Image Processing Mapping command or Image Processing Threshold command Locking sequences f necessary the active frame of a sequence can be locked as a source or destination image locking was described in the Sdecting a different sourceand destination section Chapter 3 Onlytheactiveframecan belocked you cannot lock twoframes from the same sequence You can however create images from specific frames in the sequence and lock these images or lock frames from two separate sequences You cannot use nspector s Edit Duplicate command or File Save As command toreproducean entiresequence The easiest way to create two identical sequences is
154. image The model should be created from an image with a smooth uniform background so that the contours of the object can be properly defined thereby making the operation moreeffective The operation returns the angleof the object in relation tothe model as well asthe match score between the object and the model However the coordinates of the occurrence are not returned Typical image Contours are well defined Speeding up the search 173 Speeding up the search To ensure the fastest possible search you should m Choose an appropriate model m Setthesearch speed constraint tothehighest possiblesetting for your application Set the search region to the minimum required Search time isroughly proportional totheregion searched so don t search the whole image if you don t need to Search the smallest range of angles required m Select the lowest positional accuracy that you need Set thecertainty level tothelowest reasonable value sothat the search can stop as soon as a good match is found m Search for multiple models at the same time if possible Choose the appropriate model The size of a model affects the search speed In general small models takelonger tofind than larger ones although very large models can also betime consuming I n general the optimal size is approximately 128 x 128 pixels if you are searching a large region for example most of theimage Small models arefound quickly when the search region i
155. including the edges of holes Note that when calculating the peri meter an allowanceis made for the staircase effect the pixel side of diagonally adjacent pixels is considered to be 1 414 rather than 2 For example the following blob where F represents foreground pixels has a perimeter of 14 242 n nln mnl The convex perimeter The Feret diameter Selecting blob features 187 In addition to the normal perimeter you can calculate the convex perimeter Convex Perimeter of the blobs The convex perimeter is an approximation of the perimeter around the blob s convex hull see below E Convex perimeter Normal perimeter The convex perimeter is derived from several F eret diameters of the blob Dimensions Thedimensions of a rectangular object areits length and width Most blobs however are not rectangular Therefore to get an indication of a blob s dimensions you need to look at other features such as the F eret diameter The F eret diameter of a blob is the diameter of a blob at a given angle Several F eret diameters are illustrated below Note that the angle at which the Feret diameter is taken relative to the horizontal axis is specified as a subscript to the F Y A 188 Chapter 10 Analyzing blobs Calculating different Feret diameters Long thin blobs With Inspector you can calculate the F eret diameter at 0 Feret X and 90 Fer
156. ing 57 Toaccomplish thestripesearch dicktheE dge and Stripe button alternatively use the Analysis Measurement Edge and Stripe command The Edge and Stripe dialog box is presented Click on the Stripe option not E dges Select the Multiple search option to search for multiple instances of the stripe Type3 in the Number edit field sincetherearethree pins in theimage A Oin the number or spacing edit field means that Inspector will search for any number of stripes or stripes that are separated by any amount of space respectively These settings will result in slower processing since the search algorithm will need to determine these values automatically The search algorithm is much more efficient and more robust if the expected number of stripes and their spacing in pixels are specified as opposed to being left as 0 unspecified The default auto polarity setting is suited for this example Set theinitial search box orientation toVertical since the pin legs will be in a vertical orientation Click on the Draw the Search Box button within the Edgeand Stripedialog box This search box button flashes blue 58 Chapter 2 A few tutorials Defining a search area 7 Now draw a rectangular search boxthat enclosesthe upper portion of the three pin legs The search box should be placed just below the pin s waist To accomplish this the box should have a negative horizontal offset of 50 pixels and a negative vertica
157. ing function To create a good mapping for a radial distortion requires a larger number of coordinates for example more than 30 distributed over the image Relative coordinate system A relative coordinate system is a cartesian coordinate system defined within the world coordinate system By default the relative world coordinate system is aligned with the world coordinate system H owever to get results relative to some object it can be moved anywhere within the world coordinate system and rotated by any angle Its unit of measureisthesame as the world coordinate system Therelative coordinate system is generally used to allow measurements to be made relativeto a particular object in the field of view especially when there are several such objects in the field of view For example the following is an image that was calibrated from a grid using the default settings By default the relative and world coordinate system s top left circlein this imageis 0 0 However to get results relativetothe origin of the object the relative coordinate system is at 2 1 image coordinate _ System HAALLLLITI HAAIIITTT ILILILILIL y axis Image of an object Relative coordinate system 261 world and relative coordinate systems 0 0 2 1 1 y 2 1 0 0 x axis vA xaxis 1 a ETETETT zz one x axis A T a relative coordinate a LE system oi ab CE LE oo Looe EE E y ax
158. ing on the mouse unmasks pixels Defining a mask using an existing ROI You can also use an existing ROI to mask or unmask a model To doso binarize the image you want to useas a mask that is set all mask pixels to 255 and all other pixels to 0 Activate 162 Chapter 9 Pattern matching the model s dialog box and select the Analysis Pattern Get New Mask command All corresponding mask pixels will be masked in the search model To erase the mask use the Analysis Pattern Clear Mask command Redefining a model s hot spot Thereturned coordinates of a search arethe coordinates of the model s hot spot reference position These coordinates are relativetotheorigin of theimage being searched calibrated or non calibrated By default a model s hot spot is at the model s geometric center If thereis a particular spot from which you would like results returned you can change the model s hotspot interactively by dicking on the mouse button in the Dimension tab of the Model dialog box and using the cursor to place the hotspot or by enteringtherequired X and Y coordinates in theappropriate fields For example if your model has a hole and you want to find results relative to this hole change the model s hot spot accordingly C Default model en New model reference position reference position Model Model Setting a model s search constraints 163 Setting a model s search constraints Once yo
159. ing only 10 bits must be converted to 8 or 16 bit images to be successfully supported by Inspector Finally measurements are made with sub pixel accuracy and results can be returned in pixels or real world units Steps to finding and obtaining measurements of markers 207 See the BrokenGear bas demo for a typical application involving nspector s measurement operations Steps to finding and obtaining measurements of markers You must first grab or load a target image in order to enable the measurement commands Although there are many types of measurements that can be taken with Inspector the series of steps outlined below using the Analysis Measurement E dge and Stripe command s dialog box are usually followed to find and obtain measurements of markers 1l Specify thetype of marker for which to search by enabling the Edge or Stripe radio button 2 For a multiple marker check the Multiple checkbox and set the number of occurrences of the edge or stripe in the number edit field 3 Settheprocessingarea generally referred to as the search box by clicking on the Draw the search box button and drawing the search box in thetarget image 4 Set the marker s characteristics 5 Search for the marker in thetarget image by clicking on the Find button 6 Read the results found in the measurement table 208 Chapter 11 Finding and measuring edges and stripes Multiple marker Markers edges and stripes
160. ion everywhere When the Preprocess option is enabled Inspector analyzes the shapeof the match peak produced by the model and determi nes if itis safetotry tofind peaks faster If the pattern produces a very narrow match peak an exhaustive initial search is performed The search algorithm tends to be conservative if necessary you can force fast peak finding by enabling the Fast find option The Extra candidates option allows you to set the number of extra peaks to consider Normally the search algorithm considers only a limited number of best scores as possible candidates to a match when proceeding at the most sub sampled stage You can add robustness to the algorithm by considering more candidates without compromising too 178 Chapter 9 Pattern matching heavily on search speed In addition you can usethe Coarse acceptance option on the Search tabtoset a minimum match score valid at all levels except the last level to be considered as an occurrence of the model This ensures that possible models are not discarded at lower levels yet maintains the required certainty during the final level At the last high resolution stage of the search the model is large so this stage can take a significant amount of time even though the correlation function is evaluated at only a very few points To save time you can select a high search speed using the Speed drop down list box For most models this has little effect on the score or
161. ions or any rearranging or processing of frames are completed before dosing the sequence Once a memory sequence is saved you cannot make further changes To change a saved sequence see Grabbing disk type sequences and Processi ng theframes of a sequence Grabbing disk type sequences When you grabtodisk thespecified type of sequenceis created for example AVI or TIFF and data is copied to disk from system memory during the grab Each grabbed frame is temporarily stored in memory before being copied to the disk In other words sequences grabbed to disk are double buffered Usually when grabbingtodisk morestoragespaceis available allowing you to grab longer sequences However the speed of the grab is generally slow because disk access is slow A sequence grabbed to disk is read only and the sequence file s content cannot be modified or processed in any way To make changes to a sequence the sequence must be grabbed as a memory sequence and modified before saving or copy the frames into a new memory sequence Alternatively you can createa new sequenceand insert frames intothe sequence See Creating sequences The actual transfer rate to disk depends on the fragmentation of your disk To obtain the best transfer rate check the fragmentation of your disk and then if necessary perform a disk defragmentation F or more information refer to your Windows documentation Creating sequences 249 Creating sequences Therear
162. is clicked the system is calibrated You can configure immediately in the Step 2 dialog box and open the calibrated file 29 User Manual OpenCIM Device Driver The ViewFlex Device Driver interfaces between the OpenCIM network and the Vision Machine System as a quality control device 3 ViewFlex Device Driver IDO Ele View Operation Al ele vix m Operation Mode BC Result Debug Ready t r Open CIM Messages gt From Manager lt Start gt lt Send DEVICEISREADY to manager mum Transmission lt Send DEVICEISREADY to manager TCP IP 172 16 12 6 Script Name Z ViewFlex Device Driver Each quality control test is defined as a separate process in the Machine Definition module The quality control test consists of three parts File The File is the Script File bas that contains the Program Function to be executed Program The Program is the Function from the File that returns the results of the quality control process to the OpenCIM Manager as Pass Fail or Error Fail Simulates test results only in Simulation Mode by determining the Pass Fail according to the percentage of failure that was input The ViewFlex Device Driver performs the following functions ViewFlex v 2 6 0201 Activates a test on the Vision Machine System Receives status messages from the Vision Machine System and from the OpenCIM Manager OpenCIM Messages dialog title in t
163. is yas The object s position within The relative coordinate system the real world can be moved to align it with the top left corner of the object Notethatthisimageisfor illustrative purposes only n general the object should not be placed over a grid because if the grid of circles and the object are not differentiated when performing the processing operation then erroneous results will be returned To move and or rotate the relative coordinate system of the current or all subsequent images usethe Options Calibration Relative Origin command Once you changethe origin and or orientation of the relative coordinate system world coordinates will be returned in this relative coordinate system 262 Chapter 15 Calibration Saving After you create a calibration object you can save it using the Options Calibration Save command You can choose to save the calibration object of the current image or the default calibration object used of all newly created images By default your calibration object will be saved in a MIL calibration format mca You can also save your calibration object in an Inspector 2 2 format However the Inspector 2 2 format only supports non unity aspect ratio distortion no rotation or distortion If you have a more complex calibration mapping and you choose the 2 2 format only the average X and Y scaling factors will be saved Displayed world units The position of your cursor in an imageis di
164. istinguish this defined region of the vasculature 1 Tofill in the selected ROI click the Arithmetic operation button Alternatively right click on the image and select Processing from the image s context menu 2 Select Fill with value from list box presented in the Artithmetic tab 72 Chapter 2 A few tutorials 1 Typelinthe Color Picker edit field then dick on Apply tofill the ROI with this color 2 Playback all the frames with this ROI overlayed on the sequence to see the positional changes of the main vessel 3 Click on the Pause button when the main vessel shows maximal displacement In the sequence this fame should correspond to frame 28 4 Repeat the steps previously described to define an ROI that encompasses the curvature of the entire vessel in frame 28 Collection demo Example 4 Grabbing an image 73 5 Takenoteof thecentroid position X Y for this defined ROI 6 Theresultant vector between theinitial vessel position and maximal displacement can be determined by calculating the difference in the centroid coordinates of these two defined ROIs To familiarize yourself with creating a collection in Inspector run the MakeCollection bas script provided in your software package To do so use the File Open command or Explorer to open the script file found in the BasicFiles folder The MakeCollection bas demonstration is a script which creates a sequence containing a variety of imagefi
165. ithms from the Options Preferences command When you use the Edit Palette Optimization command the palette will use the algorithm specified by the Options Preferences Display tab command For a more complete explanation on these algorithms see Computer Graphics Principles and Practicesecond edition Foley et al The Options Preferences command has an option At creation to optimize the palette upon creation or loading of an image This option is only available when the Own palette option is selected it is unchecked by default and the common palette is used to display the image Enabling this option essentially informs Inspector to find the best palette for the image when it is created or loaded by using the selected algorithm Since palette optimization requires extensive computations it is not always efficient to usethe At creation option Chapter 4 Improving image quality Ihis chapter describes how to improve the quality of your images by increasing contrast removing noise and compensating for aspect ratio distortions 104 Chapter 4 Improving image quality Image quality Prior to extracting information from an image many applications require that you obtain the best possible digital representation of that image Several factors affect the quality of an image These indude Poor contrast Often the contrast brightness attributes inanimageis poor This usually means that the full dynamic range of pixel value
166. l offset of 30 pixels from Pattern 1 s X and Y coordinates respectively The height of the box Should be 20 pixels A suitable search box with dimensions 290 pixels width x 20 pixels height is illustrated below Adjust your search box area if necessary and then click the flashing search box button a second time when finished Now dick on the Graph button to see the edge values for every edge profile valuein the search box region Example 2 Scaling calibration and pattern matching 59 Onthegraph six peaks arepresented E ach pair of adjacent peaks represents two distinct edges one positive and one negative An edge value greater than 10 is necessary for accurate edge stripe finding results 10 Click the Find button The centers of thethree stripes found by Inspector will appear as pink crosses 11 A Quick View Multiple Find Summary dialog box appears A synopsis of the stripe search results including the number and mean angle of the 3 stripes found are displayed in this dialog box Other data is also generated in calibrated units Look at the mean angle and standard deviation generated from the stripe search If the standard deviation is within the expected range you can bereasonably assured that the stripes found are valid Otherwise choose Close and redefinethe stripe find operation 60 Chapter 2 A few tutorials Quick View Multiple Find Summary x Multiple Find Number 3 Transfer
167. lay color using a minimum color distance measure between theoverlay RGB valueand the RGB values in the palette of the one band image scaled to match the buffer type if necessary Analyzing graphic objects Profiles When the Profile radio button is selected graphic objects are drawn in the overlay and once drawn the profile of the pixel values underneath the outline of the object outlined in yellow will be generated the profileis taken in the direction of the arrow If you are drawing a filled shape the graph drawn will show the pixel values underneath the edge of the object See Chapter 8 Histograms profiles and measurements for more information on profiles Analyzing graphic objects Measuring objects When the Measure check box is selected graphic objects are drawn in the overlay and once drawn a Quick View table characterizing the object s properties is generated When the Transfer check box is also selected the drawing tools become outlined in yellow and once an object is drawn its positional properties are also transferred to the current measurement table See Finding and measuring edges and stripes for more information on measuring objects Multiple mode If you plantousethesameannotation or ROI drawingtool more than once you should enable multiple mode Multiple mode keeps a tool enabled so you can draw several shapes When using this drawing mode the current tool remains selected until you select another tool
168. le such as the number of occurrences found and their mean angle Note however that if the occurrences found areclosetothe horizontal axis the mean angle result might be invalid since the angles can lie both in the first and forth quadrant For example while one angleis calculated as 1 another similar angle is calculated as 359 such that the average is skewed 222 Chapter 11 Finding and measuring edges and stripes Types of statistics Tolerances Viewing in a graph Copy to programs or Scripts Measurement statistics Inspector allows you to maintain statistics on markers and measurements Thereare fivetypes of statistics maintained X and Y positions distances angles and widths By default one set of statistics is maintained for each type However you can maintain separate sets of statistics for each type by creating statistic objects For example you can maintain separateangle statistics for the top and bottom edges in an image by creating two statistic objects of type angle You can assess measurement results against specified tolerances Settingtolerances can beuseful during automation sinceresults can beimmediately read and acted upon F or tips on determining appropriate tolerances see the Keeping analysisstatisticssection in Chapter 7 Notethat nspector also reports a quick summary of all assessments obtained for a particular marker or measurement You can view statistics in a trend or di
169. les of various filetypes and displays each file as a single thumbnail image Example 4 Grabbing an image In this example you will grab two images using a camera and an installed frame grabber digitizer Then you will perform a standard arithmetic operation to determine the pixel difference between these two grabbed images As a final exercise you will remap this small range of pixel values tothe full pixel value range for an 8 bit image 0 to 255 sothat the difference in values is visible 1 Whenaframegrabber isinstalled and a camerais attached EN the Continuous Grab tool button will be available 2 Click on the Continuous tool button to grab the required image Alternatively usethe File Grab Continuous command A live window is opened 3 Usethe Digitizer Control dialog box to make any necessary adjustments to your image The changes are displayed immediately 74 Chapter 2 A few tutorials Performing arithmetic operations on images Locking and unlocking targets 4 10 Make surethe Continuous drop down list box is set to Normal as illustrated below A A lith v Normal Double Buffered Averaged Freeze the grab on the required frame by clicking on the Halt Grab tool button alternatively usethe File Grab Halt Grab command UsetheFile Save command to savethe first image grabbed with the Normal setting as normal jpg Click on the Continuous tool button again to restore the live
170. lling the equipment check for signs of shipping damage If any damage is evident contact your freight carrier and begin appropriate claims procedures Make sure you have received all the items listed on the packing list If anything is missing contact your supplier Installing the ViewFlex System Install the ViewFlex system in the following order Assemble the camera stand and attach the camera to it Connect software protection key Install the ViewFlex software The ViewFlex installation is supplied on one CD ROM Do not plug the camera into the USB port of the computer until after the software installation Assembling the Camera Stand ViewFlex v 2 6 0201 To assemble the camera stand do the following 1 Put one of the upright posts hollow aluminum pole into the base Make sure it goes all the way in Lock the post into place by tightening the knob on the base Make sure the post does not move Option if you want to place the base on the floor connect the two upright posts together e Insert the posts together e Twist and lock the posts e Insert the post assembly into the base Lock the assembly into place as described above Open the knob on the arm unit Slide the arm unit onto the post 7 User Manual Tighten the knob Attach the camera rod to the arm unit Tighten the knob Slide the camera mount out of the Intel camera A eS Attach the camera to the camera rod with the tri
171. lly recognized trademarks and tradenames are hereby acknowledged Copyright Matrox Electronic Systems Ltd 1999 All rights reserved Disclaimer Matrox Electronic Systems Ltd reserves the right to make changes in specifications at any time and without notice The information provided by this document is believed to be accurate and reliable However no responsibility is assumed by Matrox Electronic Systems Ltd for its use nor for any infringements of patents or other rights of third parties resulting from its use No license is granted under any patents or patent rights of Matrox Electronic Systems Ltd PRINTED IN CANADA Contents Chapter 1 Getting started 05 4 21 Welcome to Inspector lesen 22 What you need to run Inspector 26 Installation me SE 26 System Allocation llle eee 29 Inspector documentation conventions 30 Menu conventions lees 30 Dialog box conventions 30 Multiple tabs llle 31 Chapter 2 A few tutorials l llli 33 Image processing sees 34 Inspector and image processing 35 Typical application steps 36 Afewexamples llle een 37 Example 1 Analyzing a blister package 38 Example 2 Scaling calibration and pattern matching e 50 Example 3 Working with many images and non standard image filetypes 62 Example 4 Grabbing
172. m if a regular pattern is expected in the Spacing edit field default is 0 meaning that spacing is not a criteria When the Spacing setting is enabled an initial search is performed tofind all edges or stripes which best conform tothe marker characteristics This group of edges or stripes arethen inspected to ensure that the spacing constraints are met Seethe Inspector BrokenGear bas demo for a typical application involving a multiple marker Measurement results 221 Measurement results If an occurence of a marker is successfully located results are automatically placed in the measurement table For an edge there aretworesult objects the Edge and Edge Position F or a stripe there are three result objects returned the two edges comprising thestripeand the Stripe Position Theedgeresult object corresponds to the line which follows the edge with the start and end points found at the boundaries of the search box while that of the Edge Position is that of the center point of this line The Stripe Position is located at the center point between both its edges and at the midpoint of the search box height The position result is optional and is only returned when the Show position also checkbox is enabled When finding multiplemarkers a Multiple Find result object is returned in addition tothe results for all occurrences found This result object contains various information concerning the group of located occurrences as a who
173. m button alternatively use the Image Statistics Histogram command The histogram graph displays the grayscale pixel values present in the defined ROI imo ci Fines D Fromtheshapeof the histogram of this small uniform area it is apparent that there is random noise in the image Using a line profile to check for sharp edges Example 1 Analyzing a blister package 43 To remove the noise in the image a smoothing operation spatial filter will be performed However applying a spatial filter might cause the edges between the background and the pill objects to lose their sharpness An easy way to keep track of edge sharpness after a processing operation is to monitor changes in the edge width and edge angle of the line profiles taken across the pills in both directions Totake these profiles you will deselect the small ROI and reactivate the original lines Then extend these lines across the pills in the horizontal and vertical direction respectively 1 Click on the original horizontal line profile 2 Create a horizontal line stretching across a row of pills in the image by dragging the line s handles across tothefinal position Extend the line as illustrated D o il 3 Reposition the vertical line profile by enabling and then dragging the line Then resize it by moving the vertical line s handles so as to extend it across a column of pills Both line profile graphs are updated with new line profile data
174. mage Erosion and dilation are discussed in Operating on your images Binary and no escape options Opening and closing 111 Tospeed up processing for opening closing erosion or dilation operations you can select the Binary or No Escape options from the Image Processing M orphology tab The No Escape option processes the operation faster by elimnating any user control over the operation such as no preview or refresh Note that when you select this option you cannot escape from the operation The Binary mode option forces processing in binary mode If you select binary modeand your imageis grayscale all non zero pixels are treated as foregound pixels and the result is white With Inspector you can customize an opening or closing operation For more information see the Creating custom structuring d ements section in Chapter 6 112 Chapter 4 Improving image quality Basic geometrical transforms I mage distortion can affect application results F or example in a medical application that analyzes blood cells if the camera does not have a one to one aspect ratio and no correction is performed the cells appear distorted and elongated and incorrect interpretation might result Dealing with distortion in such images can lead to incorrect results To resolve distortion problems you can either calibrate your imaging setup using Inspector or you physically correct your image Calibrating offers a certain advantage ov
175. marker with vertical orientation Search ini iege _ Search box i la box l Stripe 2 1 Stripe Valid Invalid Orientation The orientation specifies the angle of the edge or stripe in relation to that of the search box The initial orientation for a new search box can be set to either vertical or horizontal by togglingtheVert checkbox beforeclickingon theDraw Search Box button Vertical orientation Horizontal orientation cell Search box Search direction Search direction The orientation and direction of the search is shown by the arrow within the search box These settings can tolerate a certain amount of rotation The amount is determined by the target image placement of the search box and the marker characteristicsettings Thegreater the degree of rotation from angle of the search box the greater the chance of not finding the marker and miscalculation of characteristics such as edge strength and width If therotation istoogreat and you cannot find the marker the marker s search box should be defined with a range of angles Searching within a range of angles 211 Searching within a range of angles Inspector allows you to search for a marker within a range of angles rather than at a specific angle by internally rotating the search box during a Find operation The range is from search box angle Delta Negative to search box angle
176. mately 22 When Angle mode is enabled available through the search box context menu the Edge Strength graph shown is the best projection found for the given angular range and tolerance Weight factors When searching for a marker the relative importance weight assigned to each of the marker characteristics is crucial tothe robustness of the operation By default 5096 of the search weight is assigned to the edge strength the remaining 50 is Inside edge Markers advanced characteristics 219 equally divided among all characteristics that can have a weight factor This makes the edge strength by far the most important characteristic in the search You can override the default weight factors by checking the Enable weight mode and adjusting the appropriate valuesin the Weight by percentage section of the Edge and Stripe Settings dialog box However to better control the search itis recommended when specifying weight factors that you assign a weight factor toall the enabled characteristics which support weight factors to a total of 100 For example in a casewhere you must distinguish between two edge markers of different contrast you can specify the typical contrast of the marker to be found If you specify only this characteristic the default search algorithm will assign a 5096 weight to the edge strength and the remaining 5096 to the contrast H owever if theedge you wanttoignorehasthe higher edge strength the required
177. me the number of the current frame and the total number of frames in the sequence Thetiming of the sequence the relative time at which each frame was grabbed appears in the main panes overlay If required the timing of the frames can be exported using the Edit Export command or thescript function SegE xportF rameTimes Sequence properties The properties of a sequence can be viewed by using the Sequencel nfotab command of theView Properties command m Number of frames Indicates the number of frames in the sequence Playback rate Indicates the speed at which the sequence is played in frames per second the playback rate is adjustable Data source Indicates the type of sequence a Thumb Size Specifies the size of thethumbnails Changing the size will rebuild the thumbnails in the pane a Keep Image Ratio Keeps the pixel aspect ratio when displaying frames as square thumbnails 244 Chapter 14 Creating and manipulating sequences a In Out Shows the number of the first frame and last frame that will be played back in the sequence You can edit this playback range by typing the required frame directly in the edit fields Alternatively you can edit the playback range by right clicking on the yellow slider and dragging the ends of the slider to the required frames Notethat although thethumbnail size can be changed the actual thumbnail is not updated if the frame has been processed or modified You cannot u
178. mes To select successive frames press the Shift key while selecting them to select non successi ve frames press the Ctrl key while selecting the frames 250 Chapter 14 Creating and manipulating sequences Dragging and dropping from an existing sequence takes data but not palette information If you need totransfer the palette save the palettein the source sequenceusingthe Palette E ditor command and then load it into the new sequence again using the Palette E ditor Deleting frames in a sequence To delete frames from a sequence you must have a memory sequence Select the frame you wish to delete and use the Sequence Delete Frame command Alternatively you can select the frame s thumbnail and press the Delete key Frames from disk sequences cannot be deleted If you must remove frames from a disk sequence create a new memory sequence and drag and drop the frames of the disk sequence into it omitting unnecessary frames Saving and compressing sequences Once a memory sequence is grabbed or created you can save it toan avi or tif fileusingtheFile Save As command When saving click the Compression button to set compression parameters to compress your frames and reducethe amount of disk space needed to store them Note that compression is only supported for AVI s If you are grabbing to disk you are asked to specify the name of the filein which to save the sequence before the grab Compressing frames durin
179. mmunicates with Inspector through OLE Automation This utility uses the current image If the current image is not an FFT type image FF T exe will perform a forward transform on the image It then creates a mask image which is the same size as the transformed image and allows you to add disks spots annuli and annular sections to the mask Fast Fourier transformations 135 When you execute FF T exe a dialog box is displayed with several options You can select thefilter type for example circle or donut filter size operations to perform on the FFT image and the operations to perform on the mask such as Add Black and Smooth If you arefiltering with a mask that has sharptransitions from backgroundtoforeground Oto 1 it will produce someartifacts sometimes called ringing The Smooth option allows you to reduce ringing by blending the transition from O to 1 Operations supported on FFFtype images The following arithmetic operations are supported on FFT type images absolute value absolute difference add add with saturate copy fill with value maximum minimum subtract subtract with saturate These operations areapplied tothereal and imaginary parts Complex arithmetic such as multiply and divide is not supported Also morphological and logical operations on an FF T type image are not supported nspector cannot apply mapping or threshold operations on an FFT type image H owever it can display alive preview of thes
180. mode when a color image is viewed in component mode each component can be associated with a different palette You can change an image s display palette using the Palette Editor which is only available from the image s context menu Thefinal quality of the colors displayed depends on the number of colors available in your Windows display resolution If it has 256 colors thesettings of the Options Preferences command in the General tab will affect what is shown on the screen maging boards with a display section which use digital keying do not have a palette for displaying 8 bit unsigned images therefore you cannot use the Palette E ditor command or display a quantized image with its palette see Creating and using oneband color images The status bar will display both the pixel value and the RGB palette entry for that value whenever a non default palette is chosen 84 Chapter 3 Dealing with images Creating user defined palettes Inspector has several predefined palettes which can be used as is or can be used as a base upon which to create a user defined palette Any of a palette s 256 entries can be changed to create a user defined palette Single entries can be changed by selecting the entry and changing its values If you want to change a range of colors you can add nodes to the palette A palette nodeis a point in the palettethat interrupts the original color range Adding one node simply changes the color of
181. mpleted a match with a higher score would be found Set the certainty level in the Advanced tab of the Model dialog box Often you know that the pattern you want is unique in the image so anything that reaches the acceptance level must be the match you want therefore you can set the certainty and acceptance levels to the same value Another common case is a pattern that usually produces very good scores say above 8094 but occasionally a degraded image produces a much lower score say 5096 Obviously you must set the acceptance level to 5096 otherwise you will never get a match in the degraded image But what value is appropriate for the certainty level If you set it to 5096 you taketherisk that it will find a false match above 50 in a good image before it finds thereal match that scores 9096 A better valueis about 8096 meaning that most of thetimethe search will stop as soon as it sees the real match but in a degraded image where nothing reaches the certainty level it will takethe extra time tolook for the best match that reaches the acceptance level 166 Chapter 9 Pattern matching The positional accuracy You can set the required positional accuracy to a High typically 0 05 pixel a Medium typically 0 1 pixels a Low typically 0 2 pixels Note that the actual positional accuracy achieved depends on the quality of the model and the target image Theless accuracy you require thefaster the search will be
182. n you use the Analysis Blob New command Inspector extracts the segmentation mask from the source region and allows you specify thethreshold between foreground and background separate touching blobs using a watershed algorithm and fill the holes in the blobs H owever if your active region is difficult to segment you can segment it using some of nspector s other processing operations and then import it into blob analysis When you import this image segmentation mask it is automatically binarized using a threshold of one An imported segmentation mask cannot be segmented from within the blob settings BLOBSET dialog box f you need to use the same blob settings for several images you can easily change the source region for the segmentation mask and grayscale values or only for the grayscale values Steps to perform a blob analysis Performing blob analysis 183 Performing blob analysis When performing blob analysis you might not know which features are relevant to your application If you are trying to distinguish between two objects for example some features might show a more notable difference than others Blob analysis therefore is usually an iterative process whereby the features to calculate are added and removed until you determine which features are relevant to your application The steps to perform a blob analysis depend on the nature of your image as well as the information you need Therefore some
183. nce cameras are the most common input device we refer to input devices as such Upon starting Inspector the System Allocation dialog box asks you to select one of the digitizers it has detected If your system has only one board you can check the Don t ask me again option andthe next time you open I nspector it will open with the same board If you have more than one digitizer available refer to the section Using multipledigitizers Specifying your camera settings Before you can grab images with Inspector as discussed in Chapter 3 you must specify your camera stype To do so select a digitizer configuration format DCF file that matches your camera s type using the Options Digitizer Settings command The DCF files can be found under your MIL folder in the DRIVERS BOARD NAME DCF folder for example D MIL DRIVERS CORONA DCF In addition DCF files can be created and edited using Matrox Intellicam which is shipped with MIL and MIL Lite The digitizer settings that are required to grab from a particular camera are specified in the DCF file If necessary you can adjust these settings using the Options Digitizer Settings command You can adjust a Theinput channel s These arethechannel s from which to grab also known as the active channel s Makesurethat your camera is connected to the specified channel See Configuring the synchronization and signal channads m The grab size This is the size of thef
184. nd a text tool with which you can annotate your images The annotation drawing tools allow you to draw the following shapes Ua CN Rotated Ellipse Rotated ellipse Rectangle Ellipse Filled ellipse Rotated Ellipse sector taken Filled rectangle pd Polygon b amp Filled polygon Freehand Filled freehand Polyline lt See the on line procedures Drawing graphic objects Adding text to your images Use the Text tool to add captions to your images When this A tool is selected and you dick on the image where you want the text to appear and a dialog box appears which allows you to type in the required text and specify its font properties To modify the text double click on it to open the dialog box and change its properties Note that once you specify a set of font attributes they are automatically selected the next time you click on the text tool See the on line procedures Adding and modifying text Annotation and ROI drawing fools 227 ROI drawing tools Using the ROI drawing tools you can define ROIs with the same shapes as those possible with the annotation drawing tools In addition you can define a rotated donut toroid ROI and a magic wand ROI Many of theseshapes areillustrated in the image below rdarngie nul d t Note that some processing operations such as blob segmentation and pattern matching use the bounding rectangle of a non rectangular ROI asthe processing regi
185. nd measurement markers Palette Viewer You can select the required color from the palette viewer If the palette viewer is not visible display it with the View Palette Viewer command Clicking on a color in the palette viewer selects that color for drawing If you areusingan 8 bit display the palette viewer displays the System palette If you areusinga true color display the palette viewer displays the current image palette if one exists otherwise it displays the default palette Specifying your drawing preferences 231 To determine the RGB values of any color in the palette viewer place the cursor over the required color the RGB values are displayed at the bottom of the window s pane Eyedropper tool You can select the required color directly from the image by selecting the eyedropper tool and then dicking on the image pixel that has the required color Selecting a color using the Color dialog box You can also select a drawing color from the Color dialog box which is accessed by clicking on the color indicators next tothe eyedropper tool Outine properties You can specify the style and width of lines created with any non filled annotation tool By default a solid one pixel wide lineis selected When clicking the outline style button on the Drawing Tools toolbar the outline tab will appear where you can set its properties Outline properties do not apply to ROIs or filled graphic objects ROIs are always dra
186. nd then load this file from the required Blob Settings file You can save each dasstoits own separate file or you can save a set of classes in one file To make selecting the appropriate features more intuitive a weight for each feature is also calculated as follows w 1 2 if osx AND w 0 if o gt x where W represents the weight o represents the standard deviation x represents the mean If a feature has a weight close to or equal to 1 the values for this feature are similar for the selected blobs Similarly if a feature has a weight close to or equal to 0 the values for this feature are very different for the selected blobs Statistics and tolerances 203 Statistics and tolerances Inspector maintains a set of statistics for all calculated blob features as well as a separate set for each specified class You can view these statistics and compare results against specified tolerances Settingtolerances can beuseful during automation as results can be read and acted upon immediately To determine appropriate tolerances refer to Keeping analysis statistics in Chapter 7 To view statistics click on the Statistics button from the Settings tab of the BLOBSET dialog box or from the All Results table Alternatively use the BLOBSET dialog box s context menu then select Statistics from the displayed menu The StatBl dialog box in which statistics are displayed opens Then use the Results drop down list box to select whether
187. nga blob class usingthe bins filter or training set methods The method you select depends on several factors such as the feature distribution of the blobs in the image and the information you have about the blobs Sinceyou can automatethe classification process you nolonger have to perform this repetitive task manually which can lead to errors The general steps for classifying blobs are 1 Definethe blob classes using the bins filter or training set methods explained below Then use any combination of these classes to classify the blobs in an image 2 Set display options for the blobs in each class These settings apply only if you have chosen to display blobs using graphic objects refer to the Specifying your segmentation preferences section 3 Select whether or not toapply theclasses immediately after a count To apply the classes immediately after a count select the Apply after count option using the Option button of the Segmentation tab Otherwise you must explicitly classify the blobs in an image 4 Specify other options specific to the types bins filter or training set of the blob classes in your application 5 Classify the blobs Note that you have to perform at least one blob count operation before you can classify the blobs For a step by step procedureseethe on linetopicTo perform a blob classification You can set several options on classifying blobs and how to display classified blobs To select
188. notations Annotation and ROI drawing tools Annotation drawing tools ROI drawing tools Specifying your drawing preferences Drawing destination Analyzing graphic objects Profiles Analyzing graphic objects Measuring objects Multiple mode Selecting a drawing color Outline properties Converting graphic objects Chapter 13 Grabbing images and specifying camera settings Cameras and input devices Specifying your camera settings Grabbing i e 22 s Grab mode 00 Setting digitizer controls More advanced options LUTs and grabbing with triggers Grabbing more than 8 bits Configuring the synchronization and signal channels Using multiple digitizers Chapter 14 Creating and manipulating sequences SequetiC6eS udo seen buie ehe Loading sequences and the sequence window Sequence properties Opening compressed sequences 244 Playing back sequences 245 Grabbing sequences sls 246 Grabbing memory type sequences 247 Grabbing disk type sequences 248 Creating sequences lle 249 Deleting frames ina sequence 250 Saving and compressing sequences 250 Processing the frames of a sequence 251 Sequences and scripts
189. ntext menu and then select the Draw graphic objects in image option Toset the display of counted blobs select Properties from the BLOBSET dialog box and then select the Display tab You can set the following display options a Graphic object type Annotate blobs with the selected graphic object outline fill or mark the centroid with a cross or marker You can also choose not to draw any graphic objects Graphic object color Set thecolor for the graphi cobject type Bloblabel Select whether todisplay the blob label of the counted blobs 142 Chapter 7 Analyzing images a Class label Select whether to display the class label of the counted blobs if you have defined classes using the Classification tab refer to the Classifying blobs section Note that the Blob label option must be enabled to display the class label To return to the BLOBSET dialog box click on OK Set the display options for measurement operations by selecting the Display tab from the measurement table context menu Type Specify the type of annotation to use to identify measurement result objects It can be set to one of the following a None No graphic object Outline Outlined graphic object Color Specify the drawing color a Show object s label Select whether to display the measurement result object s label result identifier a Show object s caption Select whether to display the measurement result object s caption
190. o other applications printed or copied to the clipboard With Inspector you can alsoload or grab sequences and then play them back Sequences area set of sequential frames that can be grabbed from a camera or loaded from AVI TIFF or DICOM files You can process a frame of a sequence as you would any Inspector image In order to grab with Inspector you must have an installed Matrox frame grabber mages can be grabbed one at a time or grabbed continuously using any camera supported by your frame grabber Inspector includes a variety of point to point statistical spatial filtering Fourier transform and morphological operations You can use these operations for example to smocth accentuate or qualify images nspector supports geometric operations includinginterpolated resizing rotation and distortion correction as well as polar to rectangular unwrapping Processing can berestricted to a rectangular or non rectangular region of interest ROI Processing can also be restricted to a selected component of a color image Image calibration Pattern matching capabilities Blob analysis capabilities Measuring capabilities Welcome to Inspector 23 I mage calibration allows measurements and analysis to be carried out in real world coordinates Various calibration mappings can compensate for non uniform aspect ratios rotation perspective foreshortening and other more complex distortions Inspector
191. object which might not be the calibration object that the source image uses m When you perform a Image Advanced Geometry Correct operation on a calibrated image the new image coordinates will still bereturned in worldunits Asis expected theimage features in the destination image have the same world coordinates as they had in the source image despite the fact their pixel coordinates have changed For example after calibrating the source image below the world coordinate system of the bottom left circle is 5 6 When you correct your image the world coordinates of the bottom left circle in the corrected image are also 5 6 even though it is evident that the pixel coordinates are different for the bottom left circle in the source and corrected images Source image Corrected image 5 6 5 6 264 Chapter 15 Calibration
192. od With Inspector you can create your own filter For more information see the Creating custom filters section in Chapter 6 110 Chapter 4 Improving image quality Removing small particles Filing holes Opening and closing Another way of improving an image might be for example to remove small partides that were introduced by dust or by filling holes in objects Thesetasks can generally be accomplished with an opening or a closing operation respectively You can perform an opening or closing operation using the Image Processing Morphology command Opening and closing operations determine each pixel s value according to its geometric relationship with neighborhood pixels and as such are part of a larger group of operations known as morphological operations Besides removing small particles opening operations also break isthmuses or connections between touching objects nspector provides the Open operation to perform a basic opening operation on 3x3 neighborhoods taking all neighborhood pixels into account Closing operations are very useful in filling holes in objects however in doing so they also connect objects that are close to each other as shown below The Close operation performs a standard 3x3 dosing operation taking all neighborhood pixels into account y Note that opening is the result of eroding and then dilating an image and closing is the result of dilating and then eroding an i
193. of features filter or selection from viewing training set Inspector s measuring capabilities allow you to quickly take position distance angle and area measurements within an image You can also have Inspector locate and measure image characteristics markers such as edges and stripes pairs of edges in an image based on differences in pixel intensities 24 Chapter I Getting started Annotation capabilities Exporting results Scripting capabilities E dges and stripes are located with sub pixel accuracy You can also take measurements between positional results obtained from different image analysis operations With Inspector you can annotate your images with text lines filled shapes or outlined shapes in any specified color These can also be used as ROIs Annotation can be added directly into your images or into an overlay buffer Inspector generated data can be exported directly to any application that suports OLE automation Exporting to Microsoft Excel or Word requires no scripting If you want to export to an application other than Excel or Word you can create a custom export script module modeled on the ones provided in Inspector for Microsoft Excel or Word You can automate your applications by recording actions in a script You can easily modify the script with the help of Expression Builder The resulting script can be executed as many times as required with Inspector s Script interpreter F o
194. ols 225 You can also characterize the shape and size of any graphic object you draw and as well as the data under the graphic objects For example you can determine the minimum maximum and mean pixel value of the data under the graphic object and the profile of its outline When annotating your images you have several drawing options For example you can choose whether to draw directly in the image destructive or in its overlay non destructi ve and you can select a drawing color and linestyleof your graphic objects Notethat this chapter deals with defining ROIs with the ROI drawing tools for an in depth discussion of what ROIs are and how they are used refer to Images in general Annotation and ROI drawing tools Theannotation and ROI drawing tools are found in the Drawing Tools toolbar If this toolbar is not visible enable it with the View Drawing Tools command The tools available from this toolbar create graphic objects which can be repositioned and resized Therotated ellipseand donut can also be rotated and a sector can be taken of the rotated ellipse Note that most of the annotation and ROI drawing tools are recordable When you are recording the tools that are recordable appear in blue tools that are not recordable appear in grey Note that none of the freehand tools are recordable 226 Chapter 12 Defining ROIs and annotating images Annotation drawing tools Inspector includes several drawing tools a
195. on Seethe on line procedure Drawing ROI objects Magic wand ROIs You might want to define an ROI based on the gray or color values in an image rather than a specific area Such an ROI can be created using the magic wand tool A magic wand ROI includes all pixels that are similar within a specified range to the selected pixel value Seethe on line procedure Drawing a magic wand ROI You can alsodefinean ROI based on gray or color values using the Image Processing Threshold command and selecting the Make ROI option 228 Chapter 12 Defining ROIs and annotating images Specifying your drawing preferences You can either use I nspector s default drawing settings or specify your own drawing preferences You can specify a Drawingdestination Selects whether todraw in theimage buffer destructive or the overlay non destructive a Graphic object analysis Displays a profile of the data underneath the outline of a graphic object characterizes its properties in a Quick View table and or transfers the results tothe current measurement table a Multiple mode Keeps the current drawing tool selected until another tool is selected Outline properties Selects a line style and width a Drawing color Selects a drawing color Drawing destination You can draw graphic objects in an image s overlay or directly in the image by choosing the appropriate radio button in the Drawing Tools toolbar When the Overlay r
196. ons The advanced options that are available depend on your particular frame grabber Advanced options are similar tothe functionality of Matrox I ntellicam and are described in the Matrox I ntellicam User Guide see the inte cam hlp file The advanced options cannot be recorded or saved More advanced options LUE and grabbing with triggers Although I nspector does not directly support some of the more advanced features of your frame grabber you can always usea script to call a MIL function that does For example you can use this method for triggers and input LUTS If your Matrox digitizer supports trigger input you can grab a frame upon the occurrence of an event Your camera s DCF file specifies whether or not to perform a triggered grab and exactly how it should be carried out You can correct or precondition input data by mapping it through an input LUT when grabbing if hardware permits Although thereis nodirect support for input LUTs in Inspector you can usethesamplescript DigLU TDemo bastocall theMIL function 238 Chapter 13 Grabbing images and specifying camera settings Grabbing more than 8 bits If you have a combination of camera DCF and digitizer that supports 10 or 12 bit input Inspector will grab into a 16 bit image H owever because of the display remapping that is required only snapshot grabs are possible when grabbing into a 16 bit image Topreview thegrabat aliverate you must force thegrabinto
197. oolbar 4 Create a short horizontal linein a uniform dark region of the image by clicking at the start position and dragging across to the final position Example 1 Analyzing a blister package 41 5 Create a short vertical linein a similar fashion Draw the lines intersecting as illustrated 6 Position these two graphs generated in a tile fashion so that they are both easily visible on screen E nsure there is enough work space to view the two line profiles and the image Theline shows a relatively constant variation along its entire length minimum and maximum pixel intensity show little variation respectively from the average value Therefore the background lightingis evenly distributed in thehorizontal and vertical direction in this small region Analyzing the noise in this small region will provide valid pixel intensity results as opposed to values due to lighting artifacts 42 Chapter 2 A few tutorials Performing a histogram to analyze noise Toanalyzethe noise in the region you will takeits histogram A histogram on the background pixels will determine how the pixel values deviate from the expected value 1 Draw a small ROI enclosing the lines used for the profiles To dothis enable the ROI drawing tool Click on the ROI drawing tool button in the tool button bar Alternatively choose the rectangular ROI tool from the Drawing Tools toolbar 3 Image2 8U TTE 2 Click on the Histogra
198. or documentation conventions 31 Multiple tabs Most of Inspector s dialog boxes are actually multiple dialog boxes that is different dialog boxes overlaid one on top of the other The main dialog box is referred to as a dialog box while its internal dialog boxes or property pages are referred to as tabs The tab name appears at the top of the tab 32 Chapter I Getting started Chapter 2 A few tutorials This chapter introduces you to image processing and provides a tutorial on Inspector s processing operations 34 Chapter 2 A few tutorials Image processing Pictures or images are important sources of information for interpretation and analysis These might be images of a building undergoing renovations a planet s surface transmitted from a spacecraft plant cells magnified with a microscope or electronic circuitry Human analysis of these images or objects has inherent difficulties the visual inspection process is time consuming and is subject to inconsistent interpretations and assessments Computers on the other hand are ideal for performing some of these tasks For computers to process images the images must be numerically represented This process is known as image digitization The digitization process divides an image into a two dimensional grid of small units called pixels picture elements Each pixel has a valuethat corresponds to the intensity or color at that location in the image Each pixel in th
199. ore information to the user CT scans magnetic resonance imaging MRI and X ray type images These are usually high resolution containing up to 12 bits DICOM image and sequence files have no standard extension Inspector recognizes files of type DCM For other DICOMs you must specify Dicom medical in the Open as field in theFile Open command There are several types of DI COM formats and Inspector can read most types Many DI COM images have several types of information associated with them such as Patient Relationship Patient Identification Image type and Study Identification If you require more information about your DICOM images usethe View Dicom Information command In addition you can choose whether or not to display this information in the image s overlay by using the View Dicom Overlay command Seethe on line help for the explanations of these two dialog boxes Collection files 95 Collection files You can easily track and manage your images and sequences using collection files Collection files are visual directories that contain image files and there is no limit to the number of images in the collection provided there is sufficient memory or disk space Each collection keeps a thumbnail or file list representation of the images or sequences in the collection as well as their file name and disk location path In addition each collection keeps track of image types and sizes allowing images of different
200. ositional data will beused shortly By default you will also obtain an on screen log for all pattern matching operations To disable this option right click on the measurement table and choose Properties from the tables context menu A Measurement Table Properties dialog box is presented Choosethe Results tab and then disablethe Log results option It is important to note that for this example you perform a pattern matching operation in a source image with a square pixel aspect ratio 1 1 Accordingly it is not necessary to physically correct the image when performing the pattern matching operation However as a general rule you must physically correct an image if its distortions can significantly impact your particular processing operation For example significant perspective distortion can affect the success of a processing operation like pattern matching 56 Chapter 2 A few tutorials Search using an occurrence of a pattern as the relative origin Changing status bar display settings from calibrated units to uncalibrated units Finding stripes and angles in an image Even if your imaging setup causes your images to have a non unity aspect ratio and you want to find a model that can appear within an angular range you will haveto correct your image before performing the search In these cases after calibrating your image use the Image Advanced Geometry Correct command to correct your image using its as
201. p for the ROI tools commands You can use a user defined ROI to process or analyze only a portion of an image which reduces processing time as well as to copy a specified portion of an image tothe clipboard or for export A user defined ROI can also be set as a destination for processing and as a source or mask of blob analysis and as a source for pattern matching model definition Note the following processing restrictions of ROIs Some operations such as blob segmentation use the bounding box of a non rectangular ROI as the processing region Pattern matching operations for example will also use the bounding box of a non rectangular ROI for selecting the model pattern 88 Chapter 3 Dealing with images General processing operations except the Geometry Rotation command and Image Convert command 8 bit Quantized operations are performed on the active ROI instead of the entire image m Profiles access all pixels under the outline of the graphic object regardless of whether it is in the ROI ROI properties ROI properties are available by calling the Image ROI Properties command or from the image s context menu The properties window displays the ROI s width and height in pixels and the pixel coordinates of the ROI s origin the top left hand corner of the bounding box For rectangular or elliptical ROIs you can change these fields directly from the properties window for other ROIs you will have to move or resize
202. pod connector on the bottom of the camera Camera mount Tripod connector Camera connections 10 Place the completed assembly in a safe convenient location near the computer Camera attached to stand Do not plug the camera into the USB port of the computer until after the software installation ViewFlex v 2 8 8 User Manual 0201 Software Protection Key 1 Without the software protection key the software is programmed to run on a 30 day trial basis 2 Plug the software protection key into the printer port of your computer 3 If you also want to connect a printer to your computer plug the printer cable into the connector on the software key Software protection key Installing the ViewFlex Software Place the ViewFlex CD ROM in the drive 2 The installation procedure should begin to run automatically If it does not select StartlRun and select Install exe from the CD ROM ViewFlex Setup EN m xj intelitek 59060900 ViewF lex Image Processing Tool ViewF lex Image Quest ServeCam Exit ViewFlex Setup dialog box You must first install the Image Processing Tool module Only then can you install the ViewFlex software 3 Click on Image Processing Tool The Matrox Inspector setup will begin to run 4 Follow the directions on your screen ViewFlex v 2 6 9 User Manual 0201 ViewFlex v 2 6 0201 5 Select No I will restart my compu
203. pping from almost any grid of uniformly spaced circles However to create an accurate sub pixel mapping your physical grid should meet the following guidelines at the working resolution m Theradius of the grid s circles should range between 6 and 10 pixels m Thecenter to center distance between the grid s circles should range from 18 to 32 pixels 22 pixels recommended 258 Chapter 15 Calibration Default axes Offset and Y axis Applying the grid calibration object Theminimum distance between the edges of the circles should be 6 pixels x x 18 32 pixels i e gt 6 pixels 6 10 pixels N PA NS P m T hegrid should belargeenough tocover thearea of theimage from which you want real world results the working area m Thegrid image should have high contrast By default the circle in the top left corner of the grid image is associated to the origin 0 0 of the real world coordinate system the first column of circles is aligned with its Y axis and the first row of circles is aligned with its X axis nr 0 0 x axis PCCP RE eo PPC CTP E SPCC Cee PPC PPC Pe CTR PEE AAALALLLLLLLIJ Real world grid sanas sresasdd Grid image y axis f necessary you can set the top left circle of the grid imageto a different position within the world coordinate system The origin of the real world coordinate system does not have to be within the fi
204. r easy recall scripts can be associated with shortcut keys or buttons on the script bar or added as menu items In addition if your script has logical problems you can use I nspector s debugging capabilities to determine wherethe problems occur Inspector offers two scripting languages Basic and C Script The version of Basic used by Inspector is compatible with Microsoft Visual Basicfor Applications C Script is a scripting language similar to C Both languages allow you to extend the functionality of your application by making script calls from Inspector to functions in external DLL s Using Basic you can also use Inspector either as a dient or a server of another Windows based application that supports OLE object linking and embedding Automation Basic can also be used to develop semi autonomous scripts that support user intervention Collection files Manual overview On line help Image Quest Welcome fo Inspector 25 With Inspector you can create collection files Collections are visual directories that contain thumbnails of images They are useful for visually tracking and managing your image and sequence files You can drag and drop images and sequences into a collection from Explorer or the Find File dialog box under Windows 95 98 or Windows NT 4 0 You can also drag and drop an open Inspector image sequence into a collection From a collection you can open one or more images intothel nspector work
205. r region Note that to usethe histogram for this purpose the region should not bea saturated area otherwise you could make erroneous conclusions Inspector plots histogram values on a two dimensional graph the horizontal axis represents the intensity level while the vertical axis represents the pixel count for each intensity level Histogram window The displayed information in your histogram graph window depends on what is selected from it s drop down list box The following is alist of options available in the histogram window m f Count is selected the number of pixels at the position of the sliders is displayed m f Area is selected the sum of values between these sliders is displayed as an actual number and as a percentage of the total sum a f Accumulated is selected the sum of values to the left of these sliders is displayed as a percentage of the total sum Graphic object profile graphs Examining pixel values using profiles 149 Notethat as you changethe position of an ROI in an image the histogram window is automatically updated Examining pixel values using profiles With Inspector you can take two types of profiles m Theoutline profile of a graphic object This corresponds to the pixel values of theimage beneath the outlineof a graphic object a X and Y profile of a region This corresponds to the summation of pixel values in each column or row X or Y direction respectively When you calc
206. r both of these calibration methods you can select whether you want to apply the calibration object to the current image or to all new images If you select the latter the current calibration object is used for all newly created loaded or grabbed images You can also load a a previously saved calibration object and apply it to your current image or to all new images using the Options Calibration Load command You can load a MIL mca calibration filethat is generated by a MIL 6 0 application or Inspector 3 0 or you can load an Inspector 2 x cal calibration file When a calibration object is associated to an image that image is known as a calibrated image When you associate a calibration object to your image the image is not physically corrected and thereforeit still appears distorted However image coordinates are returned in world units you can position the cursor anywhere in the image and thecorresponding world coordinates aredisplayedin status bar of the image Also when you perform any analysis on a calibrated image your results are returned in world units You can physically correct a calibrated image by using the Image Advanced Geometry Correct command Note that Inspectors scripting functions take values only in uncalibrated units although they can return results in either calibrated or uncalibrated units Similarly it is important to mention that most dialog boxes for example measurement and 256 Chapter 15
207. r creates an 8 bit copy of the image on which to perform the operations 160 Chapter 9 Pattern matching Defining a search model Thefirst stepin performingasearch istodefinea search model There aretwo ways you can define your search model m You can manually createit by selecting an area in an existing image using the ROI drawing tool in the model s source image followed by the Analysis Pattern New Model command m You can have Inspector automatically select a unique one for you called an automodel usingtheAnalysis Pattern New Automodel command In this case nspector searches in an existing image for the most suitable unique area of a specified size and creates a model from this area Automodels are most useful in alignment applications Note the search model is always rectangular in shape When using non rectangular ROI drawing tools to define a search model the model will becreated usingthe ROI s bounding box In this case a mask might be necessary to exclude those pixels outside the non rectangular ROI SeetheM osaic scr demo for an exampleof an application which uses an automodel Saving a model Once you have set the model s search constraints you can save the model using the mod extension When you retrievethis model at a later time the Model dialog box appears with all tabs exactly as they were when the model was saved T he model does not have to be redefined or its constraints reset if a model
208. ram equalization operations change the intensity distribution histogram of your image With Inspector you can perform a uniform histogram equalization as well as an exponential and Rayleigh equalization Theuniform histogram equalization results in a more uniform distribution of your image s pixel values Thetable below lists thedensity functions used for each equalization N ote however that you donot have to be familiar with these functions to perform a histogram equalization Output probability density model Transfer functions Uniform P g 1 i Lus 8min PO 8min I Emax Emin Emin 8 8max Exponential Ca 2 47 s bir P g e min 8 88min y nl PK 8 8min Rayleigh 5 8 8 min g Emin ta 207 In oF 2 1 PKf P i 8min 2a i 9 ale a 8 28min F The cumulative probability distribution Pe by its cumulative histogram i PADo V Helm m 0 f of the input image Is approximated Refer to Digital Image Processing William K Pratt United States J ohn Wiley amp Sons 1978 p 318 For more information see the on line procedure To perform a histogram equalization operation Low pass spatial filters Rank filters Applying low pass spatial filters 109 Applying low pass spatial filters Spatial filtering operations determine each pixel s value based on its neighborhood values They allow images to be separated into high frequency and low frequen
209. rameto grab You can set the size of a grabbed frame to the size of the camera resolution by grabbing into a new window or to the size of the current window by grabbing into that window Grabbing 235 More settings are available depending on your board and camera All settings unless otherwise specified are saved in the Windows Registry For board specific information refer to the readme See the on line procedure Selecting a digitizer configuration format Grabbing With Inspector you can grab images one at a time Snapshot mode grab images continuously keeping only the last frame or grab a sequence of frames Sequences are a special case of grabbing and will be discussed later see Sequences To start a grab usethe File Grab command or dick on the appropriate grab button Snapshot Continuous Sequence on the Digitizer toolbar Grab mode In continuous mode grabbing is initiated into the window specified with the Options Preferences Digitizer tab Continuous mode is very useful when adjusting and focusing your camera Once an image has been frozen you can continue grabbing onesnapshot at atime or restart in continuous mode You can select one of three grab continuous modes from the Grab mode drop down list next totheGrab Continuous icon on the Digitizer toolbar Normal Grabs directly to the display If the depth of the image or the grabbed data is greater than 8 bits thegrab will be displayed using th
210. rate for points inside and outside the working area This mode cannot compensate for non linear distortions such as lens distortions The perspective transformation mode best fits a global perspective transformation function to the set of image coordinates and their real world equivalents Calibrating using coordinates in an image In situations where it is not possible to grab an image of a calibration grid to define a complex mapping it might still be possible to calibrate your imaging setup by using a list of pixel coordinates and their associated world coordinates Although nspector does not support this directly you can use a script that calls MIL s McalList function Use McalList to specify the list of coordinates and the calibration mode The sample script CalibGrid bas shows how to calibrate an image using a list of coordinates Thelist of pixel coordinates and their associated real world coordinates define the pixel to world mapping The more coordinates you specify the more accurate the mapping You 260 Chapter 15 Calibration use a list of coordinates to calibrate when you explicitly know the real world coordinates for a given set of pixel coordinates The specified pixel coordinates should cover the area of the image from which you want real world coordinates the working area In the case of perspective distortion knowing the world coordinates of 4 points in the image gives sufficient information to create a mapp
211. rations can be performed more efficiently Note that images with full grayscalelevels areuseful for sometasks but have redundant information for other tasks Thresholding can be used to binarize an image Binarizing reduces an image to two grayscale values for example 0 and 255 as below Notethat binary images can be used as a mask to identify blobs in a blob analysis application f ji Fo F ar t F s i prs pa Ff F VATA Cu We tae v out In the above image the objects of interest the nuclei are darker than everything else in theimage so a simple binarization will separate the objects from the background In other cases the objects might be both darker and lighter than the background To identify the objects in these cases force values above and below object values to a suitable background Thresholding images 117 value and force object values to a suitable foreground value as shown below This can be done using a band pass two level threshold Determining threshold value from histogram If you select the autothreshold check box of the Image Processing Threshold command Inspector will automatically determine the threshold value from the source image s characteristics Specifically a histogram of the source image is internally generated then the threshold value is set tothe minimum value between the two most statistically important peaks in the histogram on the assumption that these peaks
212. rayscale values you can use the Source and or Mask button in the Segmentation tab of theAnalysis Blob New command For an example see cambl ob3 scr m When you usea script to move an ROI to specific areas in an image and then usethis new region as the source For an example run the MissingBlobDemo bas Note that blob operations always needs a valid source image So you must switch tothe next source image before dosing the current source image The camblob3 scr uses a temporary imageto get around this Also notethat the new source region must be the same size as the existing one 186 Chapter 10 Analyzing blobs Selecting blob features In Inspector most features fall into one of four main groups Thearea and perimeter of a blob m The dimensions of a blob m Theshape of a blob m Thelocation of a blob The following sections discuss some of the blob features available for selection To select blob features in Inspector use the Feature tab of the Analysis Blob New command The names of specific features as they appear in the Feature tab are listed in the following sections in bold face Area and perimeter Sinceblob calculations assume a pixel is 1 unit wide and 1 unit long the area of a single pixel is equal to 1 and the perimeter is equal to 4 Thearea Area of a blob is then the number of square pixels in the blob exduding holes and the perimeter Perimeter is the total number of pixel sides along the blob edges
213. rch model 167 Preprocessing the search model Once you are ready to search for the model normal or automodel you must preprocess the model by enabling the Preprocess checkbox This ensures that for time critical applications preprocessingis doneand is performed outside of the search operation The preprocessing stage uses the known model to decide on the opti mal search strategy for subsequent search operations Preprocessing should be performed after all search constraints havebeen set To doso enablethe checkbox found on the either the Advanced tab or Angle tab of the Model dialog box If you savea model todisk after preprocessing it preprocessing changes are stored with the model it is not necessary to preprocess it again after restoring it However you must preprocess again if the following constraints are modified m Speed m Advanced tab all constraints a Angle tab all constraints m Mask m Positional accuracy 168 Chapter 9 Pattern matching Rotation By default the angle of the search is 0 However you can specify a rotational range of up to 360 as well as the required precision of theresulting angleand theinterpolation modeused for the rotated model There are two ways to search for a model that can appear at different angles Rotated model search Search for rotated versions of the model a Circular overscan model search Search for models taken from the same region in rotated imag
214. requency domain ES A combination of the three frequencies The image in frequency domain Tad aad aad Ladd aad add aad Tad call aad aad aad aad aad sad aad aad Ladd aad Ladd aad TTT dt cad cad cad cad cad Tad iad aad Ladd aad Ladd aad Tad iad aad Ladd aad Laad aad Tad call aad aad aad aad Lad Image with noise pattern Fast Fourier transformations 133 Filtering in the frequency domain After you have performed a fast Fourier transform and identified any consistent spatial pattern in your image which is noise you can create a mask to remove the noise Note that this section is written with the assumption that you are familiar with frequency domain mathematical relationships The theory behind filtering in the frequency domain is not explained in this document Tofilter constant spatial patterns using FFTs you must typically 1 Perform a forward transform which results in an FFT type image Findthefrequency components representing the noise and design a mask to remove these components Apply the mask totheFFT typeimage This directly affects the internally maintained real and imaginary components of the image Finally perform an inverse transform to obtain a filtered image in the spatial domain Frequency domain Image noise pattern removed Dots signifying noise pattern 134 Chapter 6 Advanced image processing Creating frequency filters A convolution between an image an
215. ripper This first position is called Pos 1 Click Record Position 1 39 User Manual ViewFlex v 2 8 0201 Calibration Step 1 dialog box Only along the X axis move the robot to a position on the X axis that would be greater X than Pos 1 Take a writing instrument and mark the position This second position is called Pos 2 Click Record Position 2 Now you will see in the Combo box the distance between the two positions Setting Distance on the X axis To achieve the most accurate measurement take a ruler and measure the distance between Pos 1 and Pos 2 by hand Change the distance if necessary according to the hand measurement a Click Set Distance a Click Next Step This opens the Step 2 dialog box Calibration Step 2 dialog box 9 If necessary move robot arm out of the camera s view 10 Click Snap in the Step 2 dialog box 40 User Manual 11 12 13 You will be able to recognize Pos 1 and Pos 2 in the image Click the Pen icon The Pen icon will start flashing after being engaged Your mouse arrow will become a cross Click on Pos 1 then make a dragging motion using your mouse toward Pos 2 This movement represents an increase in the X axis shows the direction of increase of the X axis Click the Pen icon again Now the X and Y axes are set and the distances are in real world millimeters When you move your mouse along the image Pos 1 becomes coordinate 0
216. rom the Windows Explorer folder into Inspector 2 Doubleclick on the sampledcm sequence file which appears as a single thumbnail imagein the collection The sampledcm sequence is loaded the 57 thumbnail frames which make up the sequence are also displayed 3 Usethescroll bar toscroll through the 57 thumbnail frames which make up the sequence 4 Now click the Play button to view the sequential series of frames 5 UsingtheROI tool button or therectangular ROI tool from the Drawing Tools toolbar draw asmall rectanglearound the blood vessel in the first sequence frame as illustrated below This is an area where the vessel is ruptured and dynamic changes in blood flow are easy to See 64 Chapter 2 A few tutorials Window leveling 4 bk m dH Hs 4 bomi haa LU 6 Click the Histogram button to get a graphical representation of the pixel intensity distribution for this particular ROI 7 Click the Play button to view the automatic and dynamic histogram update for the region selected as the blood flow through the ruptured vessel undergoes dynamic changes Now you will determine the appropriate window leveling setting to use to remap the range of pixel intensities in the source region or entire image to a new range that yields greater contrast nspector automatically sets the input range using the minimum and maximum pixel values in the source ROI Values below above the values in the input range are mapped to th
217. rranged Reactivate the sampledcm file Click on the thumbnail of frame 1 Hold down the Shift key and then click on frame 30 Using the mouse drag and drop these frames into the new sequence that you recently created Check for the small sign on the mousecursor when you dragtothenew memory Sequence window Note becausesample dcm is a disk sequence theframes are copied not moved to the open memory sequence In the case of a memory sequence you must hold down the Ctrl key when dragging frames to another sequence otherwise frames will be moved instead of copying Now dick on the Window Leveling button or usethe Image Processing Mapping Window Leveling command to return to the Mapping tab of the Processing Operations dialog box Again ensure that the Output Range is set for a full pixel range from O to 255 and adjust the Input Range to between 75 and 160 Now click on the Apply button to set this new pixel range to frame 1 This pixel range will provide optimal contrast to distinguish the vessel from the background Example 3 Working with many images and non standard image file types 69 Using the magic wand tool to define and fill a region of interest v Measure al a UN Itis worth noting that once a processing operation is applied the preview becomes unlocked automatically 10 Click on the thumbnail of frame 12 and Apply the pixel re mapping operation 11 Apply this mapping operation to th
218. s My Computer E ViewFlex for SCORBASE Program Group ViewFlex v 2 6 12 User Manual 0201 2 Software Operation ViewFlex Toolbar The ViewFlex Toolbar contains icons that are shortcuts for selecting certain commands and functions ViewFlex rf x A Gigi S v v sae 9 9 B ViewFlex Toolbar e Image Processing Tool e Camera 1 e Camera2 e ACL Terminal for ViewFlex e Abort All e Go Position e Calibration e OpenCIM Device Driver e About e Exit Image Processing Tool Open the image processing tool Matrox Inspector Refer to the Matrox Inspector Instruction Manual for more information ViewFlex v 2 8 13 User Manual 0201 YX Matrox Inspector 3 0 Eie Edit View lmage Script Help Dice be 2 oa se o at e 12 BEI e o semp lid For Help press FT AStart inbox Microsof David TY Microsoft Word Zr Image Quest Ye Matrox Inspe sane zm Matrox Inspector Camera Camera 1 Camera 2 ServeCam z When two cameras are used a Select Video Device dialog box will prompt you The camera icon opens the Camera capture window or the Client Camera window Client remote camera operation is available only in ViewFlex for SCORBASE for ER 4pc As soon as Client Camera is activated it connects to the server and displays the computer IP address where the ServeCam is located as soon as it detects it NN IET Camera
219. s PRINTLN SETPOS pos 1 60 Position pos receives the coordinate value of position in the Current measurement Table while the coordinate value of Z is 60 mm PRINTLN SETPOS pv 4 S table1 80 Position pv 4 receives the coordinate value of position 3 in the measurement table named table while the coordinate value of Z is 80 mm PRINTLN SETPOS pv ALL table2 80 22 User Manual ViewFlex v 2 6 0201 Project Vector pv receives the coordinate values of ALL positions in the measurement table named table2 while the coordinate value of Z is SO mm The following example project shows how the ACL program is used with the Vision System in order to automate the robot and the Vision System The example is made from two programs e ACL for the robot task e Visual Basic Script for the vision task GOPOS ACL Note N should be a Global Variable e The code initiates the global variable VF before sending comments to the Vision System so that it can cause the ACL program to wait until the vision command is finished This happens when VF is 1 e The program tells the Vision System to Snap then executes the GetPos function in the script The results of the GetPos function is a string that is an ACL command If the search results are 3 for example then set N to 3 set nz3 e SETPOS takes the position of the object from the measurement table that was found after the search and convert
220. s 0 to 255 is not being used s Random noise There are two main types of random noise a Gaussian noise When this type of noise is present the exact value of any given pixel is different for each grabbed image this type of noise adds to or subtracts from the actual pixel value a Salt and pepper noise also Known as impulse or shot noise This type of noise introduces pixels of arbitrary values that are generally noticeable because they are completely unrelated to the neighboring pixels Random noise can be caused for example by the camera or digitizer because in general electronic devices tend to generate noise If the images were broadcast the distance between thesending and receiving devices also magnifies the random noise problem because of interference m Systematic noise Unlike random noise this type of noise can be predicted appearing as a group of pixels that should not be part of the actual image Systematic noise can be caused for example by the camera or digitizer or by uneven lighting If the image was magnified microscopic dust particles on either the object or a camera lens can appear to be part of the image Distortions Distortions appear as geometric transforms of the actual image These can be caused for example by the position of the camera relative to the object if it is not perpendicular the curvature in the optical lenses or a non unity aspect ratio of an acquisition device Techniqu
221. s Lock Sourcelcommand For operations that require twosources such assomearithmeticoperations you must lock the second source by using the Targets Lock Source2 command A destination can also be set using the Lock Destination commands An alternative way to lock an image ROI as a second source is to select it from the S2 list box on the Selector 90 Chapter 3 Dealing with images Note that locked images appear in blue on the Selector n addition a blueicon appears in a locked image s context menu By default locks are removed after the operation is performed To keep images locked for more than one operation usethe Targets Keep Targets command Targets remain locked until you call the command again or quit I nspector To remove locks without performing the operation that is to return tothe default source and destinations usetheTargets Unlock All Targets command If you did not use the Keep Targets command for your operation and you want to use those targets again usethe Targets Restore Targets command If there were no previously locked images Restore Targets is disabled Note that all these commands are also available from the image s ROI s context menu Processing color images Unless otherwise specified processing operations are performed on all the color components of an image n some instances however it is only necessary to process one of the image components If you want to process the component in pl
222. s and positional results from the scale calibrated image in real world units To test this movethe cursor over the image and note the reported coordinated in the image s status bar It isimportant to mention that most dialog boxes for example measurement and pattern matching take positional and measurement datain uncalibrated pixel units whileall tables present results in calibrated units I nteractively you can determine the uncalibrated equivalents by setting the status bar to show uncalibrated coordinates Scripting functions which access results can access results in calibrated or uncalibrated units Now you will perform a pattern matching operation on the calibrated threepins tif source image You will first definea rectangular region to be used as the search model search pattern Then you will search the wholeimagefor occurrences of matches to your model to ensure that the three pins are present In addition the position of the first occurrence of the pattern will be used to set up the subsequent stripe finding operation Example 2 Scaling calibration and pattern matching 53 Usethe ROI tool button or select the ROI rectangle tool from the Drawing Tools toolbar to draw a region around the pin s waist The ROI s origin should begin at coordinate location 1 40 4 40 and its width w and height h should beroughly 1 50 and 0 85 respectively as illustrated bel ow nini Click on thePattern Matching button
223. s it to positions that can be sent e ALL takes every position from the Measurement Table and sets them into positions in the vector e At the end of GoPos the program moves the robot to the position in the vector PROGRAM GOPOS SET VF 0 PRINTLN SNAP WAIT VF 1 SET VF 0 PRINTLN RUN GETPOS WAIT VF 1 SET VF 0 PRINTLN SETPOS POS ALL 60 WAIT VF 1 FORI21TON MOVED POS I ENDFOR END GETPOS BAS 23 User Manual Abort All ViewFlex v 2 6 0201 Format Description e In order for the Run command to work the script must be loaded e The model for the object whose script will be searched should also be loaded e When the script will be executed a measurement table with the positions will be created The script will also send a command that initiates a sign of the number of objects that were found Function GETPOS As String IMAGE1 Insptr ImgGetCur Insptr ImgSetCurrent IMAGE1 R_Def ALL_BANDS Insptr ImgConvertType TO_8U Insptr MeasNew M X MOD Insptr PatGetCur Insptr PatSetCur M X MOD x Insptr PatFind Insptr ImgClose Num lInsptr MeasGetLastID GETPOS SET N CStr num End Function Ctrl A Immediately aborts all running programs and stops movement of axes Ctrl A is the fastest software method for stopping program execution and halting movement of all axes It can be used at any moment even while entering another command in order to instantly halt programs and
224. s not too large Adjust the search speed constraint You can adjust thealgorithm s speed As you increasethespeed the robustness of the search operation the likelihood of finding a model decreases Higher search speeds reducethe positional accuracy very slightly Effectively choose the search region and search angle You can improveperformance by not searching the wholeimage unnecessarily Search timeis roughly proportional totheregion searched set the search region target image size to the minimum required You can also improve performance by selecting the lowest positional accuracy In addition for an 174 Chapter 9 Pattern matching angular search select the smallest range required in combination with the highest Tolerance setting possible and lowest angular Accuracy setting possible Actually keeping the Accuracy option disabled is best when disabled the accuracy is equal tothe tolerance The pattem matching algorithm for advanced users Normalized grayscale correlation is widely used in industry for pattern matching applications Although in many cases you do not need to know how the search operation is performed an understanding of the algorithm can sometimes help you pick an optimal search strategy Nommalized Conelation Thecorrelation operation can be seen as a form of convolution where the pattern matching model is analogous to the convolution kernel In fact ordinary un normalized correlation is
225. s pattern matching capabilities allow you to find a pattern referred to as a model in an image with sub pixel accuracy Pattern matching can help solve machine vision problems such as alignment and inspection of objects You can createa model manually or havel nspector automatically select aunique one called an automodel from an image I n addition you can havelnspector search for all occurrences of the model or the specified number of occurrences n some cases the orientation of the model and or the target image can play a factor in the search operation Inspector includes a number of options to deal with such cases each opti mized for different image and background conditions F or example you can find the orientation of a target image or you can search for occurrences of a model at any angle in thetarget image I nspector s blob analysis capabilities allow you to identify and measure connected regions of pixels commonly known as blobs or objects within a grayscale image You can calculate upto forty selected features of these blobs such astheir area center of gravity perimeter and F eret diameter You can also automatically discard blobs that arenot of interest and classify the remaining blobs according to the values of the features The classification capabilities let you automatically classify blobs based on simple criteria such as ranges bins of one value or based on more sophisticated criteria such as combination
226. s value because its perimeter will be much larger than its convex perimeter Note that Compactness is quicker to calculate than Roughness since it is derived using only the area and perimeter 190 Chapter 10 Analyzing blobs Holes n some cases you can also distinguish blobs by the number of holes that they have Number of Holes For example you could distinguish between bolts and nuts by counting their holes However this is not a very robust measurement since a single noise pixel in the wrong place is counted as another hole Blob points Box Min X Box Min Y L gt Box Min X Y Max at X Min L gt Box Min X Box Max Y Selecting blob features 191 Finding blob locations Finding thelocation of blobs in an imagecan sometimes be more useful than finding their shape or size For example if a robotic arm needs to pick up an object it can use the location of that object in an image as a guide You can alsousethe blob location to determine if a blob touches the image borders If there are any such blobs you might want to adjust the camera s field of view so that all objects are completely represented in the image or you might just want to exclude these blobs since certain features such as the area of the blob would be misleading You can determine the following blob points X Min at Y Min Box Min Y First Point X First Point Y Box Max X Box Min Y 4 4
227. saved 198 Chapter 10 Analyzing blobs Filtering blobs based on features In addition to filtering blobs outside a specified area range you can also filter blobs based on the result of a specified feature In other words you can excludefrom theresults any blob whose specified feature is above below different or equal toa specified value In addition you can also exclude blobs whose specified feature is between or outside a specified range of values All values or ranges of values must be specified in calibrated units if image is calibrated You can select whether or not to calculate the statistics for filtered blobs Defining a filter To define a filter first make sure the required image and corresponding blob settings are loaded Then open the Blob Edit Filter dialog box in one of the following ways a Clicking on the Filters button from the Settings tab um Opening the filter file name cls from the Settings tab a Selecting Filters from the BLOBSET dialog box or All Results window s context menu m Usethe Analysis Blob Filter command In the Blob Edit Filter dialog box you can select the feature s and specify the value or range of values for the blobs to remove for the corresponding feature For a description of the elements in the Blob Edit Filter dialog box seethe on line topic Blob EditFilter Dialog Box Filter options Saving and loading filters Filtering blobs based on features 199
228. se Automatic Exposure control to have the camera automatically adjust the image according to the brightness slider setting e Use the Light Sensitivity slider to control how long the image must be too dark or too light before the camera adjusts Move the slider all the way to the left for the camera to adjust immediately Move the slider all the way to the right for the camera to adjust after 10 continuous seconds of poor exposure e Use the Brightness slider to set how bright the automatic exposure should make the scene Choose Manual Exposure control to adjust the exposure settings yourself e Use the Gain slider to make the scene dimmer or brighter e Use the Exposure Time slider to set the length of time light will hit the sensor for each image Click Restore Defaults to set the sliders to the factory settings for this tab This tab lets you control the colors in the video Using the sliders you can instantly see the effect each is having on the video picture as you move the slider Adjust the sliders until you are satisfied with the picture Saturation Drag the slider to the right to increase the intensity of the colors drag the slider to the left to decrease the intensity Moving the slider all the way to the left changes the colors to black and white Hue Drag the slider to the left to increase the blue in the video drag the slider to the right to increase the green Red Green Blue Drag the red green and blue sl
229. se coordinates relative to the top left corner of a calibrated or uncalibrated source image not the entire image and are therefore dependent on the blob s position within the source image Segmenting an image 193 Segmenting an image There are two ways to select your segmentation mask m You can either create it from your sourceimage based on a few specified settings m Usea Specific image From a source image To create a segmentation mask from a source image 1 Process your image so that blobs are as separate and distinct as possible from each other and the background 2 Select theimageas your source image by setting the option in the Mask tab of the Blob Settings Properties dialog box 3 Binarize your image from within the Segmentation tab view This tab allows you set pixel values between two selected thresholds to 0 and all other pixel values to 255 For more information see the on line topic To createa mask from the source image Importing a segmentation mask Often the image on which to perform blob analysis requires more sophisticated segmentation techniques than straightforward binarization such as separation and or filling holes n such cases you can segment your image using some of the processing operations available in Inspector and then import the segmented image into blob analysis as a mask When you specify an image as a mask the region used to create the mask is binarized automatically all gr
230. se the View View Component commands on a sequence window Opening compressed sequences The most common file types for sequences are audio video interleave AVI and TIFF AVI files use many forms of compressi on F or example an AVI filecan bea keyframed delta file or the file can be compressed using proprietary compression schemes CoDecs coder decoders are sometimes required to open files with specific compression schemes CoDecs are software modules that implement various compressi on decompressi on Schemes Some CoDecs areinstalled by default with Windows others are installed separately Within the AVI file there are identifiers which tell Inspector which CoDectouse If you save an AVI using Inspector you will beabletoreload it in Inspector but if you get an AVI from another source you might not have the right CoDec to load it unless it is in a standard Windows format For information about compressing decompressing files with the MJ PEG module see the readme file Playing back sequences 245 Playing back sequences You can view the entire sequence in forward or reverse order by pressing the appropriate Play button n addition a sequencecan beviewed repeatedly by dickingtheL oop button You can display a required frame by single stepping through the sequence using the Next or Previous buttons Alternatively you can movetheslider until the required frame appears To better view a particular frame you can zoom in or
231. sociated calibration object Note that after correction results are still returned in real world units Next you will search for and measure three pin legs in the image In this case you will usecoordinates of thefirst pin waist Pattern 1 match determined from the pattern matching operation as therelative origin for setting up a search area search box When automating this application using scripting it is important to note that Inspector s scripting functions take values only in uncalibrated units although they can return results in either calibrated or uncalibrated units It is important to remember most dialog boxes and tables do not take and return values in an identical fashion I nteractively get uncalibrated positional data in thestatus bar To doso right click on theimage and chooseProperties from theimage s context menu The Image Properties dialog box appears Select the Display tab and then set the Status Bar Format option so that Show Calibrated is not selected Now we will use nspector to check for the structural integrity of the three pin legs on the chip in our image because after much wear and tear the pin legs might be bent Sincethe pins themselves constitute a pair of edges a stripe an easy way to determine if their legs are bent is to search for three edge pairs stripes in the image and then analyzethe angle measurements calculated by Inspector Example 2 Scaling calibration and pattern match
232. space This manual is divided into two parts Part describes how to load grab process and analyze images Part describes how to automate your application and how to use Inspector as an automation client or server The appendices describe aspects of Inspector not covered in previous chapters This manual assumes you are familiar with basic Windows operations if necessary review your Windows documentation Notethat all procedures except for the ones in Chapter 2 A few Tutorials are only available in the on line help included with nspector You can obtain context sensitivehelp by pressing F Lor by using the context sensitive button on the I nspector tool bar Induded with Inspector is Image Quest a multimedia introduction to image processing that is divided intotwo parts theory and problem solving It includes text images video as well as real life applications implemented with Inspector 26 Chapter I Getting started Installing Inspector What you need to run Inspector n addition to your Inspector CD you will require a hardware key a two sided 25 pin connector to run your Inspector software once the 30 day evaluation period has expired To use Inspector you require the following m An 80486 processor or higher m Windows 95 98 or Windows NT 4 0 We recommend that you have at least 32 Mbytes of memory if you are running Inspector with a VGA card 64 Mbytes if you are using a digitizer board a
233. splayed in the status bar By default the position is displayed in calibrated units You can position the cursor anywherein your calibrated image to see corresponding calibrated units in the status bar If you do not want the status bar to display positions in calibrated units then uncheck Show Calibrated from the View Properties Display tab Retumed results M easurement and analysis results of a calibrated image are returned in calibrated units in the Quick view Measured Object Properties dialog box Measurement table T TABLE and Blob table AllRes Note that script functions that access results have the option of retrieving uncalibrated or calibrated results To retrieve calibrated results add the CALIBRATED predefined constant to functions such as MeasG Result and BlobGetResult Note that even though results can be accessed in calibrated or uncalibrated units script function parameters must be given in uncalibrated units For example the parameters for drawing functions ROIs and search boxes are in pixel units Displayed world units 263 Processing calibrated images Depending on thetype of operation performed on the calibrated image the destination image is associated with the following calibration object m When performing processing operations into new images the new image is associated with the same calibration as the source image m For geometrical operations the new image uses the default calibration
234. st data To view the entire list of resulting values in text format select View Data from the View menu or from the graph s context menu Note that this command is useful when you want to transfer text to other software packages a Move graph markers in single steps To move the markers on the graph in single steps select the required marker using the mouse and then usethearrow keystomove the markers Features of histogram profile graph windows 153 View required components To view only the required components of a color image select Visibility from the graph s context menu then deselect viewing of a particular component by clicking on that component s letter To reselect a component click on its letter again Note that displayed components have a check mark beside their letter This view allows you to compare component results between more than one histogram or profile Plot as bars or line To view the histogram profile as a bar graph select View as BAR from the View menu or from the graph s context menu I n bar view you can display each entry with at least one screen pixel ensuring that all resulting values are plotted This is useful when displaying in low resolution and when trying to figure out the drop across an edge To display each entry with at least one screen pixel in bar view select Fit 1 Value Pixel from the View menu To 9o back to line view select View as LINE fromthe View menu or from the graph s context menu
235. stribution graph These graphs were discussed in theK eepinganalysis statistics section in Chapter 6 You can copy statistics to other programs or scripts F or details see the on line help Note statistics are not maintained by default they must be enabled Chapter 12 Defining ROIs and annotating images This chapter describes how to define ROIs of different shapes and annotate an image 224 Chapter 12 Defining ROIs and annotating images Using ROIs and graphic annotations Once you have grabbed or loaded an image you might want to perform operations on localized regions of interest ROIs of the image In addition you might want toannotateyour image with labels and or drawings or to record your findings directly on theimage You might also want to manually segment an image that contains touching blobs You can usethe ROI and annotation drawingtools availablein the Drawing Tools toolbar to define regions of interest on which to perform operations or to add graphic objects to an existing image The Drawing Tools toolbar also allows you toadd text lines and filled or outlined shapes to your imagein any color The following imageshows an X ray annotated with text lines and a magic wand ROI LEGEND Medial Tubercle of the Calcanus B1 1st Metatarsal anterior B2 bth Metatarsal anterior B1 Medial longitudinal arch span B2 Lateral longitudinal arch span Annotation and ROI drawing to
236. sult table Blob calculations assume a pixel is 1 unit long and 1 unit wide If you have calibrated your source image results can be accessed in calibrated or non calibrated units When the pixel aspect ratio of an image is not 1 1 pixels are not square sotheresults of blob calculations can be misleading In this case you should either resize or calibrate the image before performing blob analysis Notethat the maximum number of blobs nspector can count is 10000 As Inspector reaches the maxi mum number of blobs that it can count it will take more time If your image contains morethan 10000 blobs define an ROI which contains less than 10000 blobs import it into blob analysis to generate the mask calculate the required information from the mask and then change the source imageand mask for another ROI in theimage You can repeat this procedure until you analyze the entire image Once you have selected features to calculate you can select the order in which to sort the feature results Using the Analysis Blob Sort command or Sort button of the Features tab the feature results can be sorted in ascending or descending order by a maximum of three features assigned as sorting keys After you have selected the sorting keys you can perform a Count operation and your blob results will be displayed sorted in the All Results table Analyzing Exporting transferring Saving and loading blob settings 197 After calculatin
237. t possible or does not correct the problem you can resize or calibrate the image 106 Chapter 4 Improving image quality Adjusting the intensity distribution From the histogram of an image you obtain a concise representation of the pixel count versus the brightness in the image commonly known as the intensity distribution The width of the occupied portion shows the dynamic range The distribution of the intensities within this range shows the image contrast Low contrast appears as a dump of pixels in one area of the image histogram medium contrast appears as a good spread from black to white high contrast appears as a bi modal histogram in which peaks exist at the outer brightness regions When an image has low contrast there might be objects that are not easily distinguished because of their similarity in intensities Inspector supports three methods of adjusting or mapping the intensity distribution m Window leveling m Adjusting the brightness and contrast linearly m Histogram equalization mapping Window leveling Window leveling takes a range of pixel values in the source image and maps these values to a different range in the destination image In essence the input range is linearly stretched or contracted to the output range By stretching the range of values actually present in an image to the full range available window leveling allows you to use all availableintensities in the image Window leveling
238. tand eee rea bedsedteaseacduarsaceatdsdesnecannacestasedsente 7 Software Imi Al 7A OM Gas iot etat pesto te oon cacao aiias as 11 Ac vg ng ViewLElex oo o epee ote EA feet Ue ence tag a nae se eee 12 2 Software Operation Lc cene rir ee enean nana mn uaaR aura auarus Ra uR 13 Vie Ele x FoolDaE Lecco oo vto oeste en e Ust facit mda mbi cen ed 13 ImapeProeessmp TOO uiuos Stange eoi as been tutu Bie eased 13 Camera Camera 1 Camera 2 ServeC am iac re m eet ssa aho co oe bu ro eden opa u 14 Video Source Camera Configuration cccceescecesnceseneeceeneeceeececeeeeeceeeeeceeeeeesreeees 16 Camera Connection Client Camera only eese 20 ACE Terminal FOr WV Ie WEG osten reo m uon pride e Gal eats 20 Fore qp pem EE 24 G TO POSION CR 25 ATT a VM EAC oz ca coc a oa oa oe ests rt C E UE 27 OpenCIM Device Driver inodo tedio e i e Lee Bi eee eae 30 3 ViewFlex for SCORBASE s 34 ViewFlex for SCORBASE L 9lbl3at oett pei ER epe tope een Pus 34 Results Dablg ss teer encore teat Oe EUH BIN QU RANK tnqe UN UO aa eus anseres eeaeee RAE Ld 34 Vision Commands in SCORBASE e eeeeeeeeseeeeeeeeeeee tenente enne enne neni 36 CaliDrattOlie edere Gusts E EET EE a E RM PAM us de M E 39 ViewFlex v 2 8 5 User Manual 0201 ViewFlex v 2 6 6 User Manual 0201 1 ViewFlex Installation Unpacking the Equipment Before insta
239. te your imaging setup in Inspector sothat theimage s pixels arescaled to real world dimensions With these scaled dimensions measurement results will be returned in real world units Subsequently you will perform a pattern matching and measurement operation to determine if the correct number of pin legs are present in the image and if they are structurally intact To determine if the correct number of pin legs are present in theimage a pattern model is used as the basis for finding similar patterns in the image Then the location of the first model will be used as the relative origin for finding the pin legs seen as stripes pairs of edges in the image nspector s measurement capabilities will determine the structural integrity of the pin legs Although in this example you find stripes in the image manually the approach taken is amenable to using nspector s automating scripting capabilities 1 Load theimage threepins tif from the IMAGES folder To do so use the File Open command or drag the file from Windows Explorer into Inspector 2 Tomagnify the region between the central and right most pin first zoom in 2 times on theimageusingthe Zoom tool button 3 Then oncethe appropriate magnification is obtained deselect the Zoom tool Calibration pixel world scaling Example 2 Scaling calibration and pattern matching 51 Now calibratethe pixel to world scaling of your imaging setup using the threepins
240. ter later when prompted at the end of the installation process Caleogi Before you can use Inspector you must restart your computer XT gt Keywords 5 Fi unctio amp Use D efinsd e DLL C Yes want to restart my computer now Je lume Se pi S Remove any disks from their drives and then click Finish to Am tying vod complete setup th EL Setup complete You only need to restart the computer after you have completed all three installations Click Finish From the ViewFlex setup click ViewFlex A dialog box will appear prompting you to select one of three options a ViewFlex ER 5 ER 9 ER 14 SV3 and Open CIM a ViewFlex for SCORBASE for ER 4pc a ViewFlex for SCORBASE for ER 4u 8 Follow the directions on the screen 10 11 12 13 14 Select No I will restart my computer later when prompted at the end of the installation process Click Finish From the ViewFlex setup dialog box select Image Quest The Image Quest installation will begin Follow the directions on the screen Restart the computer after you have completed the Image Quest installation Make sure the camera s shutter is open 10 User Manual Snap button not required in ViewFlex Shutter cover Digital PC camera 15 Connect the camera to a USB port if you are installing ViewFlex on the PC Note If you are installing ViewFlex for SCORBASE you will need to install the ViewFlex S
241. the overlay or as a profile are measured and these results are placed in the Quick View Measured Object Properties dialog box graphic objects drawn directly in an image buffer cannot be measured Once measured the graphic object s positional results can then be transferred to the measurement table and the graphic object s keypoints used for further calculations Using graphic objects provides an easy and efficient way to quickly measure various image features such as distances angles or areas Finding average intensities using graphic objects When you take quick measurements of graphic objects in an 8 bit or 16 bit unsigned image the minimum maximum and mean values the pixels under the graphic object will be calculated and placed in the Quick View Measured Object Properties dialog box Taking measurements between result objects 155 Taking measurements between result objects A variety of measurements can betaken between result objects in the measurement table All result objects in the measurement table contain keypoints Keypoints are points within a result object from which measurements can betaken Different types of result objects havedifferent sets of keypoints f you select two or three result objects in the measurement table and click on calculate only measurements that can be taken usingtheselected objects areavailable F or objects which have more than one keypoint a drop down list box allows you to c
242. the image s color space RGB and HSL true color images When you savetrue color 3 band images information about the color space RGB or HSL is not saved Therefore when Inspector loads any 3 band image it assumes that it isan RGB color image If you know that an imageis an HSL color image you should indicate this using the View Properties in the Image tab command Operating on regions of interest in your image 87 Operating on regions of interest in your image ROIsaredefined regions of images They areoften rectangular but they can have any shape You might want to perform more localized processing on an ROI An image can contain several ROIs when created each is assigned a name beginning with R The default ROI is the entire image and is named R imagename DEF Unless you lock targets only one ROI can be selected at any onetime when performing operations Even when theentireimageis theROl make sure the required ROI is selected when performing operations when selected the ROI s name will appear in the selector bar Defining ROIs To define ROIs based on areas of several different shapes or on the gray or color values in an image Usethe ROI tools on the Drawing Tools toolbar Certain types of ROIs can beresized and moved using the Image ROI Properties command See Using ROIs and graphicannotations in Chapter 12 Defining ROIs and annotating images for moreinformation about drawing tools Seethe on line hel
243. tion is a type of neighborhood operation that determines the new valuefor a pixel based on the weighted sum of the pixel and the pixel s neighboring values The weights are often represented in a matrix known asa kernel and determinethe effect of thefilter For example applying the following kernel results in a sharpening of an image 1 1 1 1 9 1 1 1 1 If none of I nspector s predefined filters suit your needs you can create your own filter or modify a predefined filter using the Image Processing Filters command and clicking on theE dit button Notethat it might be easier to simply modify a predefined filter if the filter you need closely resembles a predefined filter A filter can contain up to nine kernels A kernel can have up to seven rows and seven columns Thefinal pixel value produced by a kernel can be offset and divided by specified amounts its absolute value can then be taken After the results of each kernel are combined a final offset and divide can be applied the absolute value of this final result can then be taken If saturation is enabled final results that overflow or underflow the range of the destination image are set to the maximum or minimum value of the destination image respectively In summary the order of operation is Creating custom structuring elements 127 1 Inspector applies each kernel of the filter to the original image using the offset divide and absolute value options defined for t
244. to copy the saved sequence to another file name outside of Inspector Sequences and scripts In sequence analysis applications scripts can be useful since these applications often require performing the same set of operations on a large number of frames For example you can create a script that determines the number of frames in a sequence makes a specific frame active and then performs operations on that frame The scri pt can then be used to select and process other frames see the Tracki ngDemo bas for an example of how to do this For more information refer to the scripting chapters Chapter 15 Calibration This chapter describes how to use Inspector s calibration object 254 Chapter 15 Calibration Calibration Using Inspector you can calibrate your imaging setup so that pixel coordinates map to world coordinates Calibrating your imaging setup allows you to obtain analysis results in world units view the current cursor position in world coordinates in status bar of the images and if necessary physically correct an image s distortions When you get results in world units you automatically compensate for any distortions in an image Therefore you can get accurate results despite an image s distortions Types of distortions You can use calibration if you have one or more of the following types of distortion Non unity aspect ratio distortion Present when the X and Y axis havetwo different scale fa
245. tton to draw a line between thetwo known points and then specify the scalefactor Theline button flashes toremind you to click on it again after you havefinished drawing and adjusting the graphic object Using the same command you can also specify the origin and angle of the world coordinates system Click on the Origin Angle button and then dick anywherein theimageto select the origin 0 0 To specify an angle draw a line from the position that you want to be 0 0 in theworld coordinate system General rules for constructing a grid Calibrating your imaging setup 257 toanother point along the world X axis Notethat if you do not draw a line zero length line only the origin is set the angle remains 0 Perspective and other spatial distortions Tocorrect perspective and other spatial distortions you can use an image of a user defined grid or a list of coordinates respectively Calibration from a grid Calibration from a grid is often used when a number of images will be grabbed using the same camera position and lens configuration Using the Options Calibration New from Grid command you can specify the pixel to world mapping from an image of a user defined grid of circles and the world description of this grid You specify the number of rows and columns as well as the center to center distance between these rows and columns in real world units X spacing Y spacing You can create a pixel to world ma
246. u define a search model it is assigned a set of default search constraints You can changethese constraints using the various tabs of the Model dialog box in order to m Set the number of occurrences to find m Restrict the search to a specific region of the target image m Set thethreshold for acceptance and certainty m Set the positional accuracy m Set the search speed In most applications the above search constraints are all you need to adjust Inspector provides more advanced search contraints for cases where you want to customize the search strategy however this is not recommended unless absolutely necessary Refer tothe The pattern matching algorithm for advanced users section for details Number of matc hes With Inspector you can specify how many matches to try to find If all you need is one good match set it to one the default value and avoid unnecessary searches for further matches If a correlation has a match score abovethe certainty level it is automatically considered an occurrence default 8096 the remaining occurrences will be the best of those above the acceptance level Search region When searching for a model in an image the search region can beall or only a part of theimage Thesearch region is actually theregion in which to search for the model s hot spot therefore the search region can be smaller than the model If you have redefinedthe model s hot spot makesurethat thesearch regi on covers
247. uage for controlling and executing robot tasks ViewFlex Variable ViewFlex v 2 6 0201 Vision functions are written in Visual Basic VB script The vision script will operate the vision task To synchronize the vision and robot tasks there is the Global Variable VF After executing the Vision Function or subroutine the VF value will be 1 VF 1 SNAP 21 User Manual Format Description RUN Format Description Examples SETPOS Format Description Examples ViewFlex v 2 6 0201 SNAP Will capture an image from the camera and put it into the Image Processing Tool RUN prog Where prog is the name of a function or subroutine in the script The function returns the ACL command back to the controller PRINTLN RUN prog This executes the function or subroutine in the open VB script file If you execute the function the results will be sent to the controller The script must be loaded before it can be executed from the ACL SETPOS Position Table index Table name Z mm Where Position is a defined robot position Table index is a position of the object in a measurement table that was found in the image after searching The reserve word ALL means that if a Position is the name of a vector all the positions in the table will be set Table name is the name of the measurement table If omitted it means Current table Z is the value of the Cartesian coordinate Z in millimeter
248. ulate a profile of an object or region resulting values are plotted on a two dimensional graph Profiles of line and other graphic objects You can determinethe outline profiles of thefollowing graphic objects line angle arrow ellipse rotated ellipse or rectangle object To obtain the profile of a graphic object select Profile from the Drawing Tools toolbar then draw a graphic object The profile s graphic object is drawn in yellow Theline profile is most commonly used You can select line profile not only from the Drawing Tools bar but also using the Image Statistics Line Profile command or theLine Profile button Line profiles can be used to determine the following m Differenceinintensities between foreground and background objects a If the background illumination is uniform m f an object at an expected location is absent or present In a graphic object s profile graph the horizontal axis represents the location along the object The direction and first pixel that is plotted in thegraph is indicated by thegreen arrow on the profile s graphic object The vertical axis represents the color component intensity levels 150 Chapter 8 Histograms profiles and measurements Outine thickness To obtain averaged profile values when your image contains noise you can set the outline thickness of the graphic object Values along the specified outline are averaged over its thickness To set the outline thickness click the
249. ults in world units you automatically compensate for any distortion in an image Therefore you can get accurate results despite an image s distortions In some cases you have to physically correct a calibrated image using the Image Advanced Geometry Correct command F or more information see Calibration 144 Chapter 7 Analyzing images Exporting results UsingtheEdit Export Settings command you can export results to packages that support OLE automation such as Microsoft Excel and Microsoft Access This allows you to compare calculated data with existing data and use these results to perform statistical computations Exporting results Before you can export data you must specify the application to which data will be exported and the format of this data You need to specify the script that specifies this required information is set inthe Export property sheet of the Options Preferences command This script is called an export script module nspector provides two export modules E xportToE xcel bas and E xportToWord bas to export to E xcel and Word respectively If you want to export to an application other than E xcel or Word you can create a custom export module H owever after you specify a different export module it will beused as the export moduleof the next export operati on Creating custom scripts to export data is discussed in Chapter 20 Communicating with other applications Exporting graphics To export grap
250. umbnail frames 21 and 28 as well Notethat you can apply the mapping toa specified range of frames in your sequence using scripting To dothis you have to loop through each frame SeqSeCurrentF rame Thepositional changes of a vessel can beimportant during each cycle To verify this for this example you will define an ROI that traces the outline of a portion of the vessel in frame 1 and then you will fill this region with a distinguishing color Next you will playback the sequence and note the frames wherein thevessel re establishes its initial position and whereit shows maximal displacement from its initial position Finally you will use nspector s measurement capabilities to determine the resultant displacement for this positional change 1 Toset the reference frame dick on the first thumbnail frame 2 Enable Inspector s measurement capabilities by clicking on the Measure option under the Drawing Options section of the Drawing Tools toolbar 3 Generatean ROI outliningthemain vessel usingtheMagic Wand ROI tool This creates an ROI from connected regions with the same pixel value range To do this click on the Magic Wand button from the Drawing Tools toolbar Notethat the magic wand ROI tool also works on image previews for example of a watershed operation 4 Click on the main vessel in theimage with the magic wand cursor 5 Movethecursor over thered ROI outline which appears A dashed square box appears
251. w conversions are performed see the on line help Grayscale source image Source RGB image Source 1 specified band of an RGB image Source HSL image Source 1 specified band of an HSL image SourceBinary image Other image and sequence file types 93 Other image and sequence file types In addition tostandard image and sequence formats Inspector supports the reading of specialized image and sequence formats raw and DI COM medical formats Raw format The raw format is the most basic way of storing data data is simply stored as an array of data values Raw data can beeither asingleframeor a sequence A raw image will typically consist of an optional frame header the image data and an optional frame footer If the data is a sequence there might be an optional file header for the entire sequence as well as headers for each frame 8 16 and 32 bit grayscale and 3 band 8 bit 24 bit color data values are supported Note that if raw data is color the data is internally stored in a packed format When determining the frame size for a 3 band color image the number of bytes pixel is 3 When opening a raw image as specified in the Open as field inthe FileOpen command or any imagewhosetypel nspector cannot detect the Raw Image Information dialog box opens You can specify whether there are frame headers in the data whether the data is unsigned or not and if the image data has its bytes swapped by checking the appropriate
252. wever they do hold for long thin blobs since such blobs have approximately constant breadth along their length Note that you can also determine the ratio of the length to the breadth Elongation Compactness and roughness Selecting blob features 189 Determining the shape When trying to distinguish between blobs the shape of the blobs can be an important feature This is because two blobs can have similar sizes but different shapes because of a different number of holes curves or edges For example in the above illustration the blobs have similar sizes but can be distinguished by the shape of their holes Note that tomeasuretheshapeof theholesin the aboveillustration you would have to identify non zero pixels as blob pixels The following features will tell you something about a blob s shape a Compactness Roughness The compactness is a measure of how close pixels in the blob are to one another It is derived from the perimeter and area A circular blob is most compact and is defined to havea compactness measure of 1 0 the minimum more convoluted shapes have larger values The roughness is a measure of the unevenness or irregularity of a blob s surface It is a ratio of the perimeter to the convex perimeter of a blob The minimum roughness value is 1 0 because a blob s perimeter will always be equal to or greater than its convex perimeter A blob with many jagged edges will havea much higher roughnes
253. window Client Camera window ViewFlex v 2 8 14 User Manual 0201 Snap Grabs captures a picture and displays it in a frame OR Inserts an updated picture from the ServeCam Options Opens a dialog box use for video settings and configuration Connection available only in Client Camera Enables connection of remote cameras on a network Exit Closes the Camera window Camera Options ViewFlex v 2 6 0201 options x Video Setting Close Configuration v Snap Into current frame Image format Camera Options Always On Top Always On Top Checking this box prevents the Camera window from being hidden by other windows Snap Into Current Frame The captured image will be put into the current frame in the Image Processing Tool The part of the image that will be inserted into the frame depends on the current frame size If left unchecked a new frame will be created each time an image is captured every Snap Image Format Opens the Video Format dialog box Video Format 2 x Stream Settings Resolution Pixel Depth bits and Compression Size bytes 240176 z 1420 bd 63360 Digital Image Format OK Cancel ply Video Format dialog box 15 User Manual Configuration Opens the Video Source dialog box Video Source 2 x Advanced Pan Tilt Zoom Capture Source Main Exposure Color Image Video Input Selector
254. wn using a solid single pixel wide red outline Profile lines are always drawn in yellow Converting graphic objects The Annotations Convert command also allows you to convert graphic objects into other graphic objects Graphic annotations graphic objects used to delimit blobs measurement edge markers and ROIs can be converted into ROIs or profiles or graphic objects merged into the overlay or buffer When merging objects into the overlay or buffer you can also change the attributes color line style or filled object of the object by selecting the appropriate buttons in the dialog box When converting ROIs you can choose to keep the ROI label the R_ identifier 232 Chapter 12 Defining ROIs and annotating images TheMove button converts the graphic objects totheir new type or drawing destination The Copy button copies the graphic object s after the button is selected select the graphic object and drag to get the copy The Select All button allows you to select all the graphic objects in your image for the purpose of conversion Chapter 13 Grabbing images and specifying camera seftings This chapter describes how to grab images with your digitizer and specify camera settings using Inspector 234 Chapter 13 Grabbing images and specifying camera settings Cameras and input devices If you have an installed Matrox frame grabber digitizer you can grab from any input device supported by your frame grabber Si
255. xels represented by a given triangle edge A low sampling value will have lots of triangles and follows the variations of the image more closely but will be slower to draw and manipulate The3D image can be zoomed or rotated about each of its 3 axes Seethe on line help for the View 3D and 3D Settings commands 100 Chapter 3 Dealing with images Using palettes with an 8 bit display resolution When usingahigh or truecolor display resolution color images are always displayed in true color However with an 8 bit display driver thereareonly 256 colors out of 16 7 M available tothe display at any given moment When using an 8 bit display resolution the Windows display palette determines the set of values currently loaded into your display hardware the contents of the hardware palette can be seen using the View Palette Viewer command The Windows palette manager determines how the 256 colors are shared among the Windows desktop and the open windows The palette manager can share these colors by using one of the following methods found under the Options Preferences Display Tab command Note the Options Preferences Display tab is not available under a true color display resolution Common palette n most applications the current image askstohaveits ideal paletteloaded intothe Windows display palette The Windows palette manager responds by loading a certain number of the requested colors and approximating the others to
256. xp RD u 0 v 0 where x and y are the coordinates in the spatial domain and u and v are coordinates in the frequency domain Fast Fourier transformations 131 Inspector and F6 The destination image of a forward FFT is of type FFT It displays a logarithmic mapping of the magnitude of the image inthefrequency domain Themagnitude provides a morevisual understanding of the FFT results It is calculated as r2 7 where R and arethereal and imaginary components respectively of theimagein thefrequency domain As such the destination image is actually composed of two internally maintained floating point images one for the real component and one for the imaginary component of the transform In the status bar of the FFT image you see the values of the real and imaginary components corresponding to a pixel If necessary the real and imaginary components can be accessed with a direct call to MIL using the I mgGeRealMill D and ImgGetl magMill D scripting functions respectively Notethat the destination image is displayed with the DC component in the center Notethat if the height and width of the source image is not a power of 2 the destination image will be the next smallest power of 2 An example Thefollowing figures show single frequency images and their magnitude Single frequency images contain only one spatial frequency component that is these images can be represented with one basic 2 D pattern Therefore their
257. xt The greater the difference the larger the edge value Thefilter rejects as possible markers any edge with an edge value below 2 For best results it is recommended that only edges with edge values above 1096 be used as markers Thefilter then finds the marker by scoring each possible edge based on geometric constraints that you specify giving each characteristic a specific weight or degree of importance The edge s with the highest score is returned as the marker Markers fundamental characteristics 213 Markers fundamental characteristic s This section describes the fundamental characteristics of edge and stripe markers that are set using the dialog box of the Analysis Measurement Edge and Stripe command ot the Edge and Stripe Settings dialog box or returned as measurement results in the measurement table A stripe marker is simply a marker with two edges therefore the discussion of edge characteristics applies to each of the stripe marker s edges However certain characteristics have special attributes applicable only to stripe markers Itis recommended that you only adjust marker characteristics if Inspector cannot find the required edge or stripe using the default values You can also assign weights to certain characteristics so that somecharacteristics havegreater influencethan others Again weights should only be adjusted if Inspector cannot find the required edge or stripe with the default weight settings
258. y can contain color information For example unlike single band images color images have three dimensions X Y and a color band dimension Thereforea color image which has three bands can be represented as three single band images SeeColor images Displayingsingle band images with a palette and Creating and using one band color images Grayscaleor single band images include packed binary images 8 16 and 32 bit signed and unsigned images as well as 32 bit floating point images The value ranges for the corresponding types of grayscale images are shown below m packed binary Otol a 8 bit unsigned 0 to 255 a 8 bit signed 128 to 127 m 16 bit unsigned 0O to 64 K byte 1 m 16 bit signed 32Kbyteto 32 Kbyte 1 m 32 bit unsigned Oto4 Gbyte 1 m 32 bit signed 2 Gbyteto 2 Gbyte 1 m 32 bit float 1 17 x 10 to 3 40 x 1078 Signed images If you are performing operations that result in negative values you should useasigned imagetypeas a destination Otherwise negative values will wrap around tothe higher end of the value range Grayscale or single band images 81 Processing a grayscale image All processing and analysis operations can be performed on grayscale images H owever the pattern matching operations can only be performed on 8 bit unsigned grayscale images Inspector will automatically convert the image to the appropriate type if required Non amp bit images nspector uses deep buffers to store larg
259. y determine the position of a given pixel using theresults of this command F or example if you see a black strip along one of your image s borders this might indicate that the selected dcf s blanking width is not appropriatefor your camera UsingtheView View as text command you can determine the number of pixels by which the blanking is off Note that when viewing a color image the red component is represented by the most significant bits Thepixel coordinates areshown in theimage status bar when dragging the mouse over the text pane View 3D surfaces 99 View 3D surfaces You can display theimageas a 3 di mensional surface usingthe View View 3D command This command maps the color intensities of theimagetothe Z or depth dimension in the 3D view Therefore the lightest colors in the image will bethe highest in the3D view and thedarkest colors will bethelowest The surface can be drawn as a smooth shaded Gouraud shaded surface of triangular patches or four sided polygons or as a set of points or as a wire frame grid You can control how the 3D imageis drawn usingthe View 3D settings command Display quality of 3D view in 256 color display resolution is poor duetothe use of dithering therefore a true color display is recommended when using the View View 3D command The 3D image is drawn with triangular patches or 4 sided pol ygons thesizeof which is determined by thesampling value The sampling value is the number of pi
260. y resolution less Chapter 4 Improving image qudlity Image quality oone trea Ea NEE EE EREA Techniques to improve images Adjusting the intensity distribution Window leveling lees Adjusting the brightness and contrast linearly 040 Histogram equalization mapping Applying low pass spatial filters Opening and closing cresie sra sia soras Basic geometrical transforms Scaling eere wanian gue Gee He eee Re wg Rotating images 0 Translating and changing the symmetry Interpolation modes Chapter 5 Image transformations Operating on your images Thresholding images 4 Accentuating edges ero oro ora wai Edge enhancers lessen Edge detection arrere s rinrinta epee en Arithmetic with images Erosion and dilation Chapter 6 Advanced image processing 125 Advanced image processing Creating custom filters Creating custom structuring elements Fast Fourier transformations FFT define s v RR RR Inspector and FFTs Anexample llle Filtering in the frequency domain Operations supported on FFT typeimages lees Polar to rectangular and rectangular to polar transform Wat

100299-d ViewFlex-v28 full 02-01

Contents

Download Pdf Manuals

Related Search

Related Contents