Home

MIDAS 2010 User Manual

1. 4 Getting Started 4 1 Data Requirements Before any data can be loaded using MIDAS 2010 the user must save the data as a tab delimited text file with extension txt The data saved in a text 10 file must be a matrix with m n dimensions The first column of the ma trix must contain the variables such as wavenumbers while the remain ing columns contain samples like absorbance units The dynamic vari able such as temperature or time is incremented by a constant amount A between each column starting from column 2 to n That is each row con tains the wavenumber in the first cell and the various absorbance data are in the remaining cells Temperature or time is incremented between each column starting with temperature or time T 0 in column 2 T 6 in column 3 and T n 2 x in column n As an example for a temperature or time dependent IR spectra mean ing that the dynamic variable is temperature or time the first column should contain the wavenumbers and the remaining columns should con tain the corresponding absorbance units The absorbance values in the cells represent the average absorbance over the temperature or time in crement since temperature or time is augmented by a constant amount 6 between columns 2 through n Visually the data matrix should look simi lar to Table 1 1400 2 0 000103 0 000257 0 000090 1398 3 0 000198 0 000513 0 000222 1396 4 0 000097 0 00036
2. 7 Analytical Tools The following plotting options are available to customize the method in which the data is analyzed and viewed Some options are strictly available to Plots 1 and 2 7 1 Selecting Data 7 1 1 Start and End X Y The drop down menus in this section allow the user to define the X and Y plotting ranges Start X and End X correspond to the start and end dy namic variables as in temperature or time while Start Y and End Y corre spond to the start and end wavenumbers Recall that the X Y and Z axes represent the temperature or time wavenumber and absorbance units respectively from the spectra plot as specified when the data was initially loaded with the loading window The Start X and End X values in the drop down menus take into ac 20 Compare Traces x 10 Time vs Absorbance Units 1 5 1 oo E 2 05 o o c c i S H SE N 0 5 1 U 420 840 1260 1680 X Time s Time vs Absorbance Units Select Data Data Setup Plot Setup gt Select Y 1390 59 ca Remove Static Co 0 Average Scale Regular 4 Mean Sp e 3 Kernel LLL S ee Min Signat o SUD A T E Apply Pre Fitter Apply Post Filter nd 4 Qoo01 End X 2085 Design Pre Filter Design Post Filter X Ticks 6 x 10 Wavenumber vs Absorbance Units IS B e 1 o o np 2 S 2 H FA 0 5 1 1359 73 1307 66 1255 58 1203 51 1151 43 1099 36 Y Wavenumber cm 1 Wavenumber vs Absorban
3. Data Cursor 61 map Only one colour map can be selected at a time The Jet colour map is designed my Matlab and covers most colours in the spectrum The Red Blue colour map is specially defined by MIDAS 2010 Its purpose it to help the user easily differentiate positive from neg ative values on a plot When a graph is plotted everything above zero is shown in red while everything below zero is shown in blue on Plots A and B Figures 23 and 24 illustrate the Red Blue colour map option 11 3 Change Title Selecting the Change Title tool allows the user to change the title of the active plot see Section 11 1 1 for the definition of the active plot The New Title window shown in Figure 25 will appear to assist the user change the plot title Since Plots A and B have the option of plotting four different types of plots namely the synchronous asynchronous phase and spectra graphs the plot title will not remain each time the plot is refreshed to a different type of plot Plots 1 and 2 are cross sections of the spectra plots Thus when a new title is given to these plots it will remain every time the plot is refreshed 11 4 Change Labels and Units Selecting the Change Labels tool allows the user to change the axes labels and units The Labels and Units window shown in Figure 26 will appear 62 TES File Insert DEAD OUER 9 CENE BS Apply Selection to All M x Synchronous vum Sn o SNE s CEST ler
4. 6 e x lt ompae Tri TES S Ex EAS number vs Absorbance Units s T c x10 1 e v emm lt IDE 2 d 7 8 ny vers ON A S 567 lt J V gt f Type Synchro lt E z gt A n i 13578 1313 44 126908 1224 72 Y Wavenumber cm 1 vs Absorbance Unts Data Setup Remove Static Co Average Apply Post Fiter 1319 2 Y Wavenumber cm 1 1363 59 1363 59 XWavenumber cm 1 X Ticks 6 Y Ticks d Figure 18 Apply Selection to All applies these ranges to all plots in the main interface That is the ranges for Plots B 1 and 2 are changed to the ranges selected in Plot A Then all plots are refreshed while applying the current analytical and viewing settings of each plot The data is processed according to the steps outlined in Section 9 10 4 Compare Traces Button The Compare Traces buttons are only available for Plots 1 and 2 They are tools enabling the user to compare more than one trace at a time When Compare Traces is pressed the respective plot is refreshed while applying 54 File Edit View Insert Tools Desktop Window Help D de r 850984 2 05 80 SAVE x 10 Time vs Absorbance Units E 1 Time vs Absorbance Unit mew MI Time vs Absorbance Units 1390 59 11384 8039 11373 2318 PEJE r Select Data Select Y 13732318 w Start X D End X 2085 U r Data Setup 5 Remove Static Comp 2 5 v 3 2 S a 5 2 EI ES N Mean Sp
5. Selecting the Apply Pre Filter or Apply Post Filter option will bring up the filtering interface shown in Figure 7 When the Apply Filter option is se lected the current or last filter applied to the data is shown on the inter face The filters are applied to the data in the Fourier domain 32 8 1 Periods in Time and Fourier Domains The period in the Fourier domain along the U and V axes are equal to the number of samples in the X and Y directions of the input data in the time domain respectively If the number of samples are nx and ny respectively then the range nx 2 to nx 2 on the U axis comprises one period in the U direction while the range ny 2 to ny 2 on the V axis comprises one period in the V direction 1 8 2 Frequency Domain Filtering The filters are applied to the data in the Fourier domain The following list enumerates the basic steps 2 taken to achieve filtering in the Fourier domain 1 Typical discrete Fourier transforms DFT are infinitely periodic This causes problems when multiplying DFT s together If periods over lap the product is subject to wraparound error Padding both the data f x y and the filter h x y ensures they both have the same periods in the frequency domain If f x y and h x y are of size A B and C D respectively we can form two padded functions both of size P Q by padding f and h with zeros Wraparound error is avoided by choosing P gt A C 1 Q gt B D 1 T
6. Sessions The Save file menu item allows the user to save figures and current ses sions This will save the current workspace figure and the mat file that is needed to save the data and work done to each plot area In order to re open the an existing file MIDAS 2010 must locate the fig and mat files so it is important that both the fig and mat files are saved A file can only be re opened if the workspace is saved The saving window shown will appear to assist the user save the figure appropriately 12 1 3 Exporting Data MIDAS 2010 can export the data used to plot graphs Plot A B 1 or 2 After plotting the desired graph the user can export the data to an Excel spreadsheet by selecting Export XLS or to a tab delimited text file by select ing Export TXT from the file menu Data exported into an Excel spread sheet is sent to a Ale file while data exported into a tab delimited text file is sent to a txt file The data used to graph Plots A and B are matrices Therefore the data sent to the Excel file are also a matrices The exported data will have the 66 Y axis down the rows and the X axis across the columns That is the front corner of the mesh graph is at top left hand corner of the matrix in the spreadsheet When exporting the matrices to graph Plots A and B into a text file MIDAS 2010 exports the plot as shown on the main interface That is to say if the current plot is the synchronous plot MIDAS 2010 exports the syn
7. esea e doe ra a 66 12 1 4 MIDAS Manual la na 68 TALO OUI S Seam gr dod ae dol i ined e E 68 12 2 Exporting Data to XLS 24 ia dudes EE 69 12 2 1 Information Worksheet 69 12 2 2 All Spectra Worksheet suh eta mh Y ees 69 12 2 3 Selected Spectra Worksheet 2 005 edo ea 70 12 2 4 Wavenumber and Time Worksheets 70 12 2 5 Synchronous Asynchronous and Phase Worksheets 71 12 2 6 Static Component Worksheet 71 12 2 7 Pre and Post Filter Worksheets 71 122 Insert Ment seu Fer dead EUR oS cR rc Eee 72 12 9 T Blanket Pileta gore O UTR OW ac 73 72 1X4 Change Miles oo 24 SEALS ALES AS kie eed 79 12 5 Change Labels and Units a Dedo eg Tt dece 74 13 Dealing with Problems 74 13 1 fspecial Error Messages o ooo o 74 13 2 Compare Traces Property Editor 19 9 Plottine Problem gt c eta moea A eo See 13 4 Stopping Matlab s Computation 13 2 FROZEN MOUSES ae coed ae S Sea oc te EE 13 6 Fro z n Computer DIT 4 X d ot V ed a eate 13 7 Contact Information clle 14 Acknowledgments List of Figures MATLAB Window lei ga eoa eere t gt Ans Browse For Folder Window MIDAS 2010 Window io dee dtes at A Load a File Interface rs Mair Window s s Ss Tn eto add Tar Td ra T wie dede de DIOS 2 urs scary Wace Fale Wace oh ae oh argo hae halon a Filter Interface Lowpass Gaussian
8. in Section 7 3 3 Fast Fourier Transform and Correlation Using the data matrix result ing from the Special Scales step MIDAS 2010 takes the fast Fourier transform of the data along the dynamic variable and then performs the univariate correlation The resulting correlation matrix is then multiplied by 2 to correct to the number of degrees of freedom 4 Synchronous Component The Synchronous component is obtained by taking the real part of the resulting correlation matrix Asynchronous Component The Asynchronous component is obtained by taking the imaginary part of the resulting correlation matrix 51 Phase Component The Phase component is obtained by comput ing the phase between the real and imaginary parts of the resulting correlation matrix asyn syn arctan 9 Plotting The viewing options are now taken into consideration MI DAS 2010 checks how the use wants to view the data matrix either Mesh Contour Mesh Contour 3D Contour Contour Fill or Waterfall The program plots the data and marks the axes with the appropriate number of tick marks The desired colour map is also applied see Section 11 2 Spectra Note that the Special Scales and Fast Fourier Transform and Correlation steps are not applied to the spectra data Rather the re maining data following the Choose Selected Data Ranges step is plot ted In broad terms the data is processed in the order in which the analytical
9. labels and units specified with the loading window 11 5 Help and User s Guide Pressing the MIDAS User Manual button will automatically open this user guide as a Portable Document Format PDF 12 Menu Items There are two menu items on the top left hand corner of the main window in MIDAS 2010 64 Insert New Ctrl N el m B ox Lae y Open Ctrl 0 F gt x Select Data Apply Selection to All ES Z Save Workspace Ctrl S Synchronous x 10 Start 0 Export XLS pr End X 12085 Export TXT a Start Y 1363 58 MIDAS Manual Ctrl M End Y 1136 00 Quit MIDAS Ctrl Q Data Setup ISE 9 Remove Static E 154 MF Steady t E B ed Apply Pre Filte E E i Design Pre F E DIS O Average E j 1 Kernel E D Apply Post Filt 5 E Figure 27 File Menu e File e Insert They are discussed in detail in the following sections 12 1 File Menu The File menu is shown in Figure 27 12 1 1 New and Open The New and Open file menu items operate similarly to the New File and Open File buttons on the toolbar To load a new set of data in MIDAS 65 2010 select New The loading window in Figure 4 will appear To open previously saved work select Open A dialogue box will appear asking the user to find and select the workspace figure Once the figure is selected the workspace window will appear as it was last saved 12 1 2 Saving Current Work
10. option MIDAS 2010 selects all the raw data originally from the input text file and does nothing to it 2 Apply Pre Filter MIDAS 2010 checks if the user selected the Apply Pre Filter option and if so the program determines which filter was last designed by the user using the Pre Filter Interface The pre filter is applied to the data following the Remove Static Component step Re call from Section 8 that the filter is applied to the data in the Fourier domain The filter application steps are described in detail in Section 8 2 3 Compute Average MIDAS 2010 checks if the user selected the Aver 49 age option and if so the program determines which average type the user wants to apply either Kernel or Dynamic Average The averaging technique is applied to the data following the Apply Pre Filter step The averaging techniques are described in detail in Section 7 2 3 Re call that if the user selects Dynamic Average with e gt 1 the resulting data matrix will be compressed along the dynamic variable Thus if the original data matrix has dimensions a b namely b points corre sponding to the dynamic variable and a dynamic average is taken every e points then the resulting data matrix has dimensions a b e Apply Post Filter MIDAS 2010 checks if the user selected the Apply Post Filter option and if so the program determines which filter was last designed by the user using the Post Filter Interface The post fil
11. plot from MIDAS main interface is advantageous it grants the user access to Matlab s regular tools Matlab treats this new figure as any other figure That is to say while the plot remains the active figure the user can enter commands in Matlab s Command Window to further alter the plot Additionally since the plot is isolated from MIDAS main interface it can be saved exported printed etcetera as desired 10 4 1 Property Editor Once the conjoining trace plot appears the user is free to modify the line colours line styles line widths marker types marker sizes and marker colours for each trace To do this select the Edit Plot icon looks like a mouse arrow on the trace plot to edit the plot Then double click on a trace and the Property Editor will appear at the bottom the the trace plot The user is free to change any of the line or marker properties at will Note that the if the user changes the Plot Type and refreshes the plot Matlab is likely to complain because MIDAS 2010 did not account for a change in 56 X Data Source auto Y Data Source Z Data Source gt 10 Marker none 3 0 SE EES Select Data Select Y 13732318 v Sank 0 Endx 2005 e Data Setup Remove Static Comp Mean Spe Apply Pre Fiter Apply Post Filter Plot Setup Scale Regular Min Signet 0 0001 XTicks e Figure 20 Compare Traces Property E
12. sum of the squared intensities for each resolution element gives consistent results with statistical correlation 4 29 05 1190 01 1230 51 1271 01 1311 51 Y Wavenurnber cm 1 Synchronous 1352 02 1311 51 1352 02 Apply Selection to All 1190 01 1230 51 1271 01 X Wavenumber cm 1 Start x End x Start Y End Y Select Data E 1880 13520161 aen Data Setup Remove Static Comp Steady State v 7 Apply Pre Fiter Average 1 ema 7 Apply Post Fiter Type Vew Scale x Ticks Mn Signal n Pos Synchronous Mesh Resolution D Reguler 0 0001 6 nos Plot Setup g 1190 01 1230 51 1271 01 1311 51 YlWavenumber cm 1 Asynchronous 185202 13 1311 51 152 02 1190 01 1230 51 1271 01 X Wavenumber cm 1 x1 Start y End x Son v End Y Select Data E 1880 13520161 em Remove Static Comp Steady State El Apply Pre Fiter Design Pi Average 1 ena 7 Apply Post Fiter Asynchronous Mesh erger 40 Seale Regular Mn Sionat vom D 6 YTicks Data Setup Figure 9 Mesh Plots Synchronous and Asynchronous Plots Y Wavenumber cm 1 119001 123051 127101 131151 135202 Je p pose E D ENANOS n BRE ESE 1352 02 1271 01 1230 51 1190
13. 01 X Wavenumber cm 1 Select Data Stat Ta End 4960 Stat 13520161 Endy mae Remove Static Comp Steady State E Apply Pre Fiter n Pre Fiter Average 1 Kema 7 Apply Post Fiter Te Synetronous Vew contour Resolution E Seale Regular Mn Signal Dm X Ticks 6 Y Ticks Data Setup WEG 6 Y Wavenumber cm 1 135202 Asynchronous 132 02 SE 1311 51 1271 01 1230 51 X Wavenurrber cm 1 Start x End x Son v End Y Scale vie 7 Apply Post Fiter Mn Signal Select Data s 1980 13520161 11514317 Steady State 7 Apply Pre Fiter Average Asynchronous Contour Resolution E Regular om 5 Ticks Data Setup lt Remove Static Comp Figure 10 Contour Plots Synchronous and Asynchronous Plots 30 Essentially when the variance scale option is selected the maximum absolute value across the dynamic variable temperature or time is found and compared to the Minimum Signal value If the maximum absolute value is found to be larger than the minimum signal value then the data across the dynamic variable for the particular wavenumber is scaled by the Euclidean length Range Range scaling divides each function by the range as defined by the maximum function value minus the minimum function value This method of scaling has little effect on the synchronous and asynchronous co
14. 5 0 000189 Table 1 Example Data Matrix 11 4 2 Step to Saving Data Matrices A simple way to save the data matrix in the text file with the appropriate requirements is to 1 Open the Report Window in OPUS The Report Window can be ac cessed through the Window tab menu Window gt gt New Report Window 2 Select and copy all the data If there is a lot of data there may not be enough memory space to copy all the data at once In this case the data must be copied in intervals Press and hold the SHIFT key to ease selecting multiple rows of data at once 3 Past the copied data into a text file Notepad is the suggested text ed itor Do not modify the pasted data It will look very messy because the columns and rows will no longer be properly aligned Although this is not too appealing to us the computer knows which data is in each row and column The data should already be tab delimitated this should be apparent 4 Save the text file in an appropriate location Make sure the file is saved with the txf extension 5 Open Matlab and set the MIDAS2010 folder to the Current Directory The Browse For Folder button has three small dots on it and is located 12 File Edit Debug Desktop Window Help ele m 5 E curent Directo ry H My Document its MIDAS2010 Shortcuts 2 Howto Add 2 What s New irent Directory Workspace Da F m B v gt New to MATLAB Watch this Vi
15. Mid Infrared Data Analysis Software 2010 A Matlab Interface for FTIR Spectroscopy Analysis User s Guide Elise Normand January 10 2011 Contents 1 Purpose 2 System Requirements 3 Installations 4 Getting Started 4 1 Data Requirements EE br A AS 4 2 Step to Saving Data Matrices 5 Loading Data Interface 5 1 Selecting a Data Bes ote ada air dede 5 2 Labeling Axes and Units ba EE po JOSE E AAA A dena nho ts 6 Main Interface 61 PlotsAandB 6 2 Plots Land 2 eu REA A e Beds ks GSS 7 Analytical Tools TL Selecting Data EE 7 1 31 StartandEnd X Y o a and Xe made dta E 7 2 Data Setup riada AS RAPS Rae Wa Rae SB EES 10 10 12 15 15 17 17 17 19 19 7 2 1 Remove Static Component 22 7 2 2 Apply Pre and Post Filtet 9s 24 aa AJAVOIOPE ar a pev wea 25 Pl ik cared uc d aa creto eod det ed ee PROP caede toa 27 Poth Type nie IR AAA AS ue deut 3 27 EE 28 7 3 3 Scale and Minimum Signal 29 Jd AXES DORS wg BRA PE aaa 32 PIO LDRGeVO SOOY E ito a il ede o tie e 32 8 Filtering Interface 32 8 1 Periods in Time and Fourier Domains 33 8 2 Frequency Domain Filtering 3939349 ee eos s 33 8 3 Plot o onies auis acate Scan ade are ah ee amem aes 34 633 ke Filter Type ranri Ree EU E UTE AAA 34 8 32 VIEW us S eae wk eR A ek a 35 Sd Filtering DultonBa ei ita a eee ERU du 36 8 4 1 Parameters Ver
16. Pre Filter 0 B me cre Q Blanket Post Filter PLC C Apply Selection to All Change Title Synchronous Change Labels and Units E P P x 10 2o orrellation Intensities Figure 28 Insert Menu Select Data Start X 0 End X 2085 Start Y 1363 58 End Y 1136 00 Data Setup Remove Static Steady Apply Pre Filte Design Pre F Average 1 Kernel Apply Post Filt filter before applying it to the data Notice how the filter type and its design parameters are listed at the top of the worksheet 12 3 Insert Menu The Insert menu is shown in Figure 28 12 3 1 Blanket Filters The Blanket Filter menu item will bring up the filtering window shown in Figure 7 The operations of the filtering window are explained in Section 8 above When a filter is selected from the Blanket Filter menu item the 72 designed filter is applied to all four plot areas in the main window There fore Plots A B 1 and 2 are refreshed in that order with the newly filtered graphs If Blanket Filter is selected from the Insert menu and no filter is de signed meaning the user pressed the X on the top right hand corner of the filtering interface then no filter is applied to the data Notice how the Ap ply Filter checkboxes are not checked and Plots A B 1 and 2 are refreshed without applying the filter 12 4 Change Title Selecting the Change Title menu item all
17. Remove Static C Average I TE Select Y ZE e Scale Re Scale Regular y 1140023 e Means em e been z 0 i y A Min Signal 0 0001 StetX 0 E Apply Pre Fiter Apply Poster MM Sionat 0 0007 0 0 2 0 4 0 6 0 8 X Ticks 6 Y Ticks 6 Enax 2085 Design Pre Fiter_ DesignPost Fiter x Ticks 6 r Select Data StetX 0 1 Compare Traces E 1 EndX 2085 EA y Start Y 14002335 lt 09r A EndY 10993569 y 08 Data Setup oo Remove Static Co ar TED el Tl Appi Reas Plot 2 0 6 J Plot B n 0 5 L1 keng E Apply Post Fiter 02 04 on P 03 0 1 1 n f 1 1 D 01 02 03 04 05 06 07 08 09 1 02 Wavenumber vs Absorbance Units Select Data Data Setup S P Plot Setup oi GEN bes Ej Select x o Remove Static C Average Scale Regular A rco Dreem MeanS y Kemet Min Signat 0 0001 0 Min Signat 0 0001 t 140023 E apply Pre Fiter E Apply Post Fiter Reverse Y 0 02 04 0 6 0 8 X Ticks 6 Y Ticks 6 End Y 1099 35 lt Design Pre Fiter Design Post Fiter v Ticks 6 Figure 5 Main Window 18 6 1 Plots A and B Plots A and B are the top and bottom left hand plots in the main win dow and function identically The purpose of these similar plots is to ease comparing the same data set while varying the analytical tools Altering the method in which the data is processed or analyz
18. Setup Bandpass Fitter Fitter Type Bandpass Cut in 0 25 Create an bandpass filter View Mesh Cutott 02 Cutin isthe first freqency defining the band pass where 0 Cut in 0 5 Resolution 60 Order 1 X Ticks 6 Cutoff is the second fregency defining the band pass where d 0 lt Cutoff lt 0 5 YTicks 6 Order isthe order of the filter The higher the order the sharper the transition Note that the order must be an integer gt 1 Fourier Domain Filter Figure 16 Filter Interface Bandpass 46 8 11 1 Alternative Designs Using the Bandpass Filter Due to the nature of the bandpass filter a Lowpass Gaussian filter can be designed if cut in 0 cutoff gt 0 and order 1 Similarly a Highpass Gaussian filter can be designed if cut in gt 0 cutoff 0 and order 1 When cut in gt cutoff a bandpass filter is designed When cut in lt cutoff a bandstop or notch filter is designed 8 12 Highboost Filter When the highboost filter is selected the window in Figure 17 is displayed The user can design a highboost butterworth filter 6 using the following variables Cutoff The cutoff variable is the second frequency defining the band pass where 0 lt cutoff lt 0 5 Order Order is the order of the filter The higher the order the sharper the transition The order must be an integer greater than 1 Boost The boost variable is the ratio that high frequency values are bo
19. a unique name Please see Section 12 2 for details outlining the worksheets created when exporting to XLS Matlab will issue Warnings saying it added a specified worksheet to the Excel file if the worksheet does not already exist This is completely normal The last message will indicate if the exporting is done gt gt Done exporting to XLS 12 1 4 MIDAS Manual Selecting the MIDAS Manual menu item will automatically open this user guide as a Portable Document Format PDF 12 1 5 Quit Before quitting MIDAS from the file menu the user has the option of sav ing the current workspace A dialogue box will appear asking if the user would like to save the workspace Pressing the X on the top right hand side of the main window will also present the dialogue box 68 12 2 Exporting Data to XLS 12 2 1 Information Worksheet The Information worksheet shows the metadata at the time the export was issued The labels and units correspond to the axes labels and units specified with the loading window DeltaX is the constant A temperature or time increment between the columns of the data matrix as explained in Section 5 3 Delta Y is the increment between the rows of the data matrix which is assumed to be constant The StartX EndX StartY and EndY are the settings of the selected data at the time the export was issued The settings of the analytical tools are also listed The filepath and other path names indicate the name of the d
20. ass Butterworth Filter Filter Type Highpass Butterwo e Cutoft 0 25 Create a rotationally symmetric highpass butterworth filter View Mesh H Order 2 Cutoff isthe cutoff frequency of the fiter and must lie within the range Oto 0 5 Resolution 60 ane Ta Order is the order of the fitter The higher the order the sharper the transition Note that the order must be an integer 1 Y Ticks 6 Select this Filter Fourier Domain Filter Figure 14 Filter Interface Highpass Butterworth 43 PltSeup Highpass Elliptical Fitter Fitter Type Highpass Elliptical CutoffM 025 Create an highpass elliptical filter 2 View Mesh H Cutottm 05 CutoffM is the cutoff freqency on the major axis where A D lt CutoffM lt lt 1 Resolution 60 Order 3 X Ticks Cutoffm is the cutoff fregency on the minor axis where 3 Alpha 0 lt Cutoffm lt 1 Y Ticks Xoffset Order is the order of the filter The higher the order the sharper Voffset the transition Note that the order must be an integer gt 1 Alpha lt an angle in radians which rotates the filter View this Filter counter clockwise through alpha Xoffset offsets the filter in the x direction An offset of 0 corresponds to the center and an offset of 1 corresponds to the edge of the filter A positive offset shifts the filter in the positive direction Yoffset offsets the filter in the y direction An offset of O correspon
21. ata file and its location within the computer 12 2 2 All Spectra Worksheet The content on the All Spectra worksheet is what MIDAS 2010 read as in put from the provided text file The first column of the matrix contains the variables such as wavenumbers while the remaining columns contain samples like absorbance units The dynamic variable such as tempera ture or time is incremented by a constant amount between each column starting from column 2 to n That is each row contains the wavenum 69 ber in the first cell and the various absorbance data are in the remaining cells Temperature or time is incremented between each column starting with temperature or time T 0 in column 2 T in column 3 and T n 2 x in column n 12 2 3 Selected Spectra Worksheet The content on the Selected Spectra worksheet takes the selection ranges into account by selecting the spectrum data ranging between StartX and EndX and StartY and EndY inclusively The variables such as wavenum bers are no longer in the first column The entire matrix contain the se lected samples like absorbance units The dynamic variable such as tem perature or time is incremented by a constant amount 6 between each column starting from column 1 to n Notice how Startx EndX StartY and EndY are listed at the top of the worksheet 12 2 4 Wavenumber and Time Worksheets The column vector on the Wavenumber worksheet shows the selected
22. ce Units Select Data Data Setup Plot Setup Select X 0 SIL Average Scale Regular y A Mean Sp e 3 Kernel e MEOCERGSEBSI nnnm Start Y 136358 v unc apre I 7 e AR j Apply Pre Filter _ Apply Post Filter Reverse Y End Y 1099 35 SA Design Pre Filter Design Post Filter Y Ticks 6 Figure 6 Plots 1 and 2 21 count from the loading window Recall that delta is the constant tem perature or time increment between the columns of the data matrix The Start Y and End Y values in the drop down menus are taken directly from the input data matrix it is the first column from the text file 7 1 2 Select Y and X Recall that Plots 1 and 2 plot orthogonal cross sections of the spectra data Plot 1 plots the temperature or time versus absorbance units along a spe cific energy it graphs X versus Z for a selected Y Therefore the user can select the specific energy using the Select Y drop down menu Similarly Plot 2 plots the wavenumber versus the absorbance units along a specific temperature or time it graphs Y versus Z for a selected X The user can select the specific temperature or time using the Select X drop down menu 7 2 Data Setup 7 2 1 Remove Static Component There are two types of static components the user can remove from their data the mean spectrum and steady state component Mean Spectrum The mean spectrum is defined as the average
23. chronous data within the selected data ranges If the current plot is the spectra the program exports the spectra within the selected data ranges The first column in the text file corresponds to the variables like wavenumbers and the rest of the matrix is the data used to plot the graph ie the synchronous or spectra matrix Since the data for Plots 1 and 2 are simple cross sections of the averaged data the data is only a set of vectors and not a matrix Therefore the data sent to the Excel sheet is two columns one contains the X and the other contains the Y data points for the plot Note that the data sent to a text tile will look very messy because the columns and rows will no longer be properly aligned Although this is not too appealing to us the computer knows what data is in each row and in which column The data should already be tab delimitated this should be apparent Do not modify the exported data in the text file Important Note When exporting to a file that already exists the com puter will ask the user if he or she would like to replace the existing file If the user chooses to replace the existing file note that the program does 67 not erase the file completely before writing to it Rather it writes over top the old file but does not erase the existing data Therefore the user will end up with a mix of data overlapping between old and newly exported data To avoid this mishap be sure to export to a new file with
24. d number to assure the window has a center cell The kernel coefficients are modified to take an average over the dynamic variable temperature or time only MIDAS 2010 uses a Matlab function called fspecial which belongs to the Image Processing Toolbox to design the kernel fspecial creates an average filter using ones 1 lxe The kernels with e 3 and e 5 are shown in Tables 2 and 3 respectively Dynamic Average When dynamic averaging is selected the mean be tween data points along the dynamic variable temperature or time is Description is from Matlab Help 25 0 3333 0 3333 0 3333 Table 2 Kernel Average e 3 0 2 0 2 0 2 0 2 0 2 Table 3 Kernel Average e 5 taken If Average is set to some number e for example then the average of the first e data points in a given row is taken and placed in a cell The following e data points from the same row are then averaged and placed in the following cell This process is repeated until the end of the row is reached or until there are less than e data points left in the row and an average cannot be taken Thus if the original data matrix has dimensions a b namely b points corresponding to the dynamic variable and a dy namic average is taken every points then the resulting data matrix has dimensions a b e Of course an average of 1 does not modify the original matrix A visual example of dynamic av
25. deo see Demos or read Getting Starte gt gt MIDAS2010 Figure 1 MATLAB Window on the top of the Matlab window it is shown in Figure 2 At this point MIDAS 2010 is ready to analyze the data in the text file MIDAS 2010 can now be used to analyze the spectroscopy data To run the software type MIDAS2010 case sensitive in Matlab s Command Window gt gt MIDAS2010 The MIDAS 2010 window shown in Figure 3 will appear Note that Mat lab will show a busy signal on the lower left hand corner of the Matlab window whenever it is computing an operation 13 Browse For Folder Select a new directory WE Desktop A Libraries B Elise Normand Computer I Sk Network Folder Computer Make New Folder E mipas2010 Mid Infrared Data Analysis Software 2010 Load a New File Open an Existing Workspace Created by Elise Normand Canadian Light Source Mid Infrared Spectromicroscopy elise normand usask ca Figure 3 MIDAS 2010 Window 14 There are two ways to pull up an interface using MIDAS 2010 The first involves loading a new data file such as a data text file and the second involves opening a previously saved workspace To load a new set of data press the Load a New File button The loading window shown in Figure 4 will appear To open previously saved work press the Open an Existing Workspace button A dialogue box will appear asking the user to find and se
26. ditor plot type Notice how the selected trace is identified by its Display Name 10 4 2 SAVE Button Pressing the SAVE button allows the user to save the conjoining trace plot as an image or as a Portable Document Format PDF The saving inter face comes to view and assists the user with the saving process The trace figure can also be saved using the save button from the trace figure itself 10 4 3 Export Toolbar There are two exporting options available to the user The user can export the traces listed in the Traces list either as a tab delimited text file or to an Excel file The file is saved in a location specified by the user 57 Es Select Data Start X 0 End X 2085 Start Y 1363 58 8 End Y 1136 00 Data Setup 6 Remove Static Co Steady 4 Apply Pre Fitter 2 Average Apply Post Filter Figure 21 MIDAS 2010 Toolbar 11 Toolbar MIDAS 2010 main interface is equipped with a toolbar where some tools are similar to Matlab s familiar tools and others are unique to MIDAS The toolbar is identified in Figure 21 11 1 General Tools The first eight tools on the toolbar are considered general because they are similar to Matlab s familiar tools The first three are the New File Open File and Save File As New File When the New File tool is selected the loading window shown in Figure 4 appears and assists the user in selecting another file to load for analysis Open File Wh
27. ds to the center and an offset of 1 corresponds to the edge of the filter A positive offset shifts the filter in the positive direction Fourier Domain Filter Figure 15 Filter Interface Highpass Elliptical 44 range 0 lt cutoff lt 1 Order Order is the order of the filter The higher the order the sharper the transition The order must be an integer greater than 1 Alpha The variable alpha is an angle in radians which rotates the filter counter clockwise through alpha Obviously and angle of zero does not rotate the filter Xoffset and Yoffset The Xoffset and Yoffset variables offset the filter in the x and y directions respectively An offset of O corresponds to the center of the filter and an offset of 1 corresponds to the edge of the filter A positive offset shifts the filter in the positive direction while a negative offset shifts the filter in the negative direction 8 11 Bandpass Filter When the bandpass filter is selected the window in Figure 16 is displayed The user can design a bandpass butterworth filter 5 using the following variables Cut in The cut in variable is the first frequency defining the band pass where 0 lt cut in lt 0 5 Cutoff The cutoff variable is the second frequency defining the band pass where 0 lt cutoff lt 0 5 Order Order is the order of the filter The higher the order the sharper the transition The order must be an integer greater than 1 45 Plot
28. e v 7 Apply Pre Fitter Design Pre Fiiter O Average 3 Kernel E Apply Post Filter Design Post Filter Plot Setup Scale Regular Min Signal 0 0001 X Ticks 6 Figure 19 Compare Traces the current analytical and viewing settings However rather than being plotted in the appropriate location on MIDAS main interface the data is plotted in a separate figure which comes to view The data is processed according to the steps outlined in Section 9 This new figure works in conjunction with the trace interface which also comes to view Notice how the trace interface is nearly identical to the Plots 1 and 2 sections on the main interface The essential difference is that the trace interface includes a new Traces sections Using the Select Y or Select X drop down lists from the Select Data section the user can select which traces she or he would like to compare Once the trace is selected press the Add 55 button to add the selected trace to the list Traces can easily be removed in a similar manner After selecting the trace from the Traces list press the Remove button to remove the trace from the list Pressing the PLOT button will plot or refresh the conjoining plot while applying the current analytical and viewing settings All the traces listed in the Traces list will be plotted on the conjoining plot The data is pro cessed according to the steps outlined in Section 9 Separating the trace
29. ed and comparing the results can be advantageous for the user The primary plots used in MIDAS 2010 are Plots A and B They have the option of plotting the synchronous asynchronous phase or spectra graphs To procure these graphs the Fourier transform and 2D correlation of the data matrix are computed and result in a matrix of complex numbers The synchronous graph is obtained from the real components while the asyn chronous is obtained from the imaginary components The phase between the real and imaginary components forms the phase graph The spectra graph plots the original data before the Fourier transform and 2D corre lation are performed As such the X Y and Z axes represent the tem perature or time wavenumber and absorbance units respectively as specified when the data was initially loaded with the loading window 6 2 Plots 1 and 2 Plots 1 and 2 plot orthogonal cross sections of the spectra data Plot 1 is the top right hand plot and Plot 2 is the lower right hand plot in the main window Plot 1 graphs the cross section along a specified wavenumber 19 That is it plots the temperature or time versus absorbance units along a specific energy it graphs X versus Z for a selected Y Plot 2 graphs the per pendicular cross section which is along a specified temperature or time That is it plots the wavenumber versus the absorbance units along a spe cific temperature or time it graphs Y versus Z for a selected X
30. en the Open File tool is selected the user can select and open an existing figure as it was last saved This tools allows users to 58 re open previously saved work for further analysis Save File As When the Save File As tool is selected the user can save the current figure and workspace in a specified location This tools al lows users to save a current work session so that it can be re openned later on for further analysis Note that two files are saved when this tool is used filename fig and filename mat are two files needed to restore previ ously saved work Both files are required when the current work session is re openned at a later date 11 1 1 Tools For the Active Plot The next set of tools assist the user in modifying the view of the active plot A plot is active as long as it is the last one plotted by Matlab That is when the Apply Selection to All or Blanket Filters see Section 12 3 1 are used Plot 2 is the active plot because MIDAS 2010 refreshes the plots on the main interface in the following order Plot A Plot B Plot 1 then Plot 2 If any particular plot is refreshed using the PLOT button then it is now the active plot Zoom In and Zoom Out The Zoom tools are identical to Matlab s zoom tools Since zooming in and out can be difficult to see within the main interface it is suggested to press the Figure button first to plot the active plot in it s own figure independent of the main interface Double click
31. eraging is shown in Figure 8 MIDAS 2010 re computes the temperature or time increment between successive points in the data matrix using 6 which was defined in the load ing window and in Section 5 3 If the user selects Dynamic Average with e gt 1 then the resulting data matrix is compressed along the dynamic temperature or time variable the increment between the columns of the data matrix is no longer Rather the increment between the columns of the compressed data matrix is now 6 x e The lists in the Start X and End X 26 2 2 3 3 Resulting Data 5 6667 7 6667 Figure 8 Dynamic Average Example drop down menus are modified accordingly 7 3 Plot Setup 7 3 1 Type There are four types of graphs for Plots A and B as previously mentioned e Synchronous e Asynchronous e Phase e Spectra 27 To procure these graphs the Fourier transform and 2D correlation of the data matrix are computed and result in a matrix of complex numbers The synchronous graph is obtained from the real components while the asyn chronous is obtained from the imaginary components The phase between the real and imaginary components forms the phase graph The spectra graph plots the original data before the Fourier transform and 2D corre lation are performed As such the X Y and Z axes represent the tem perature or time wavenumber and absorbance units respectively as specified when the data was initially loaded wi
32. ff 11 2 Colour Maps The selected colour map applies to both Plots A and B Every time a new colour map is selected Plots A and B are refreshed with the new colour 60 mx fs e Time vs Absorbance Units sty 1365517 y _ 05 A Mens Jl gt 0 t E Apply Pre Fter E 7 d 05 5 E ei PostFaer sj E Design Post Fier 15 L n 0 420 840 1260 1680 x Time s Time vs Absorbance Unts Select Data Data Setup na Setup zer ENPIDE CE COMO MED gt ens lt CHE ge 1365 52 State d der Mi Sinat oam 136552 131923 127294 1226 65 1180 36 cere l 7 Apply Pre Fiter ELT xw 6 X Wavenumber cm 1 E o sync en nchronous x10 Wayenumber vs Absorbance Units 1587 v 11350021 y 4 7 X Wavenumber 1201 5778 ss S ps Y Wavenumber 12305082 3 E Z Conr Henger 7014e 008 El Apply Pre Fiter 3 E Design Fre 8 Average EE s e ES J E 7 Apply Post Fater 05 gt Pla Setup E i type Smero im 132116 127487 Cam 118229 1136 Y Wavenumber cm 1 Mae Wavenumber vs Absorbance Unts 1226 65 Resolution Select Data Data Setup lt Plot Setup 127294 ondes 5 Scale Regum elll Select 0 e FE LS i Scale bedis v ES 180823 Min Soret 00091 Stet Y 1385517 WE See EE EECH Y Wavenumber cm 1 138552 1365 52 qwavenumber cm 1 xke 6 Vins len en Al O SE E ea Figure 22 Toolbar
33. herefore obtain the padding parameters P and Q 2 Compute the two dimensional fast Fourier transform of the data with the padding to obtain the DFT of the data 3 Multiply the DFT by the Fourier domain filter H x y G H x DFT 4 Obtain the real part of the two dimensional inverse discrete Fourier transform of the resulting product G x y The real part g x y is now in the time domain due to the inverse Fourier transform 5 Crop the result to the original size to eliminate extra padding 8 3 Plot Setup The following plotting options are available to customize the method in which the filter is viewed 8 3 1 Filter Type There are eight types of filters the user can apply to the data e Lowpass Gaussian Filter e Lowpass Butterworth Filter 34 Lowpass Elliptical Filter Highpass Gaussian Filter Highpass Butterworth Filter Highpass Elliptical Filter Bandpass Filter Highboost Filter These filters are explained in the following sections 8 3 2 View There are six ways to view the desired filter e Mesh Wireframe parametric surface graded with colour e Contour Two dimensional isopleth or topographic map with con tour levels graded with colour e Mesh and Contour Wireframe parametric surface with a contour plot beneath the mesh both are graded with colour e 3D Contour Three dimensional contour plot graded with colour e Contour Function Filled two dimensional contour plot using c
34. herefore you cannot use filtering or kernel average options since these options use the fspecial function 13 2 Compare Traces Property Editor Note that the if the user changes the Plot Type and refreshes the plot Mat lab is likely to complain because MIDAS 2010 did not account for a change 74 in plot type Close both the trace interface and conjoining plot to stop Mat lab s complaints 13 3 Plotting Problem If the user notices MIDAS 2010 incorrectly graphed a plot simply refresh the plot by pressing the PLOT button It is possible that MIDAS encoun tered a problem because of the particular selected plotting options If a warning is issued on Matlab s Command Window please record it as well as the current plotting options The programmer should be able to find and repair the problem 13 4 Stopping Matlab s Computation Sometimes MIDAS 2010 might take a long time to generate or refresh a plot This might be because the selected data set is large or because Matlab encountered a problem To stop Matlab s computations press Ctrl C in Matlab s Command Window A warning is most likely to appear because Matlab was forced to stop in the middle of an operation 13 5 Frozen Mouse If for some reason the user does not have a mouse or if the mouse has frozen MIDAS 2010 can still be manipulated with the keyboard By press ing the Tab button the user can jump between options The Spacebar is 75 used as a select
35. in Figure 7 is displayed MIDAS 2010 uses a Matlab function called fspecial which be longs to the Image Processing Toolbox This function returns a rotationally symmetric Gaussian lowpass filter with standard deviation c positive fspecial creates a Gaussian filter using ni ni hy n1 n2 GE h ni Ta h n na bam uf hg Sigma The variable sigma is the standard deviation c 8 6 Lowpass Butterworth Filter When the lowpass butterowrth filter is selected the window in Figure 11 is displayed The user can design a rotationally symmetric lowpass but terworth filter 8 using the following variables Cutoff The cutoff variable is the cutoff frequency of the filter and must lie within the range 0 to 0 5 Order The variable order is the order of the filter The higher the order the sharper the transition The order must be an integer greater than 1 Description is from Matlab Help 37 Plot Setup Lowpass Butterworth Filter Filter Type Lowpass Butterw Cutoff 025 Create a rotationally symmetric lowpass butterworth filter View Mesh Order 3 Cutoff is the cutoff frequency of the fiter and must lie within the range Oto 0 5 Resolution 6 X Ticks 6 Order isthe order of the filter The higher the order the sharper i the transition Note that the order must be an integer gt 1 Y Ticks 6 View this Fiter Select this Filter Fourier Domain Filter Figure 11 Filter Interface Lowpass B
36. ing on the plot while the tool is activated resets the plot to its original view Selecting the same tool from the menu will turn the selected tool off 59 Pan The Pan tool is identical to Matlab s pan tools Since panning can be difficult to see within the main interface it is suggested to press the Figure button first to plot the active plot in it s own figure independent of the main interface Double clicking on the plot while the tool is activated resets the plot to its original view Selecting the same tool from the menu will turn the selected tool off Rotate 3D The Rotate 3D tool is identical to Matlab s rotate 3D tools Since rotating can be difficult to see within the main interface it is sug gested to press the Figure button first to plot the active plot in it s own fig ure independent of the main interface Double clicking on the plot while the tool is activated resets the plot to its original view Selecting the same tool from the menu will turn the selected tool off Data Cursor The Data Cursor tool is identical to Matlab s data cursor tool The X Y and Z axes refer to the temperature or time wavenumber and absorbance units axes respectively from the spectra plot as defined in the loading window Thus when the synchronous asynchronous and phase plots are viewed the Y axis is listed twice since these plots have wavnumbers plotted along two axes Selecting the same tool from the menu will turn the selected tool o
37. lect the workspace figure Once the figure is selected the workspace window will appear as it was last saved 5 Loading Data Interface The loading window named LoadGUI_2010 appears upon request from the MIDAS 2010 window Note that only one loading interface can appear at a time meaning only one text file can be loaded into MIDAS 2010 at a timet 5 1 Selecting a Data File A file must be selected before any other adjustments can be made When the Select File button is pressed a dialog box will appear asking the user to select the desired text file After the file is selected the file path is shown on the loading window Mn future versions of MIDAS the programmer should allow for more than one file to be loaded at a time Check out singleton guide and property inspector in Matlab 15 Load a File Select a File File Path none m Axes Labels Axes Units X axis Time X axis Y axis Wavenumber Y axis Zaxis Absorbance Units Zaxis Points Setup Delta 15 s Time or temperature increment between each data point Load this File Figure 4 Load a File Interface 16 5 2 Labeling Axes and Units The X Y and Z axes corresponding to the spectra plot can be named at this stage as well as their corresponding units When plotted the X Y and Z axes represent the temperature or time wavenumber and absorbance unit
38. line at the Canadian Light Source CLS at the University of Saskatchewan Please refer to the CLS website for specifics detailing the end stations techniques and optics for the Mid IR beamline http midir lightsource ca MIDAS 2010 can be used as a verification tool The current software program on the Mid IR beamline used to analyze the data from the Bruker FTIR spectrometer and Hyperion microscope occasionally encounters some plotting problems MIDAS 2010 was created in an effort to avoid these plotting issues as well as to ease selecting specific parts of the data for analysis Aside from some of the regular Matlab tools and utilities MIDAS 2010 can help the user load and prepare the data compute the fast Fourier transform and cross correlation of the selected data and graphically dis play the results in a variety of ways Additionally MIDAS 2010 has the ability of applying various types of filters to the data as to eliminate high frequency noise MIDAS 2010 is the second version of this program the first of which was simply named MIDAS 2 System Requirements In order for MIDAS 2010 to work properly the following software must be installed on the user s computer e Matlab Version 7 6 0 R2008a Must include the Image Processing Toolbox Note that MIDAS 2010 has not been tested using previous versions of Matlab e Any text editor such as Notepad Version 5 1 Service Pack 3 e OPUS Version 6 0 or better 3 Ins
39. ls Dynamic Average Example io A a VD 0 d O oO A Q N Mesh Plots Synchronous and Asynchronous Plots o Contour Plots Synchronous and Asynchronous Plots Koch ked Filter Interface Lowpass Butterworth aaa aaa N Filter Interface Lowpass Elliptical C2 Filter Interface Highpass Gaussian 6 14 Filter Interface Highpass Butterworth 43 15 Filter Interface Highpass Elliptical 44 16 Filter Interface Bandpass ua RW RSS REA 46 17 Filter Interface Highboost aromas aa a 48 18 Apply Selection to EE E 54 19 Compare Traces V 2p 4 ridad 55 20 Compare Traces Property Editor 26 4 td aoe a 57 21 MIDAS 2010 Toolbar s iud rdg do ag E ad 58 22 Toolbar Data Cursor bobos oe nie oet one de eda 61 23 Red and Blue Colour Map ugeet AE a 63 24 Contour and Rotated Red and Blue Colour Map 63 25 New Title Window 65420 d A a ae a Gage BS 63 26 Labels and Units Window less 64 24 File Ment ose aus utn dus Lui uie dd E eel ue dod e ue te e 65 28 Insert Menu a so 3 4 6 E Pk 4E DRED LG eoe d 72 List of Tables 1 Example Data Matrix 2844s XE E Se eed 11 25 KernelAverage e S35 veux qoe Y ORI EY v des 26 0 KemelAyeraget e 9o asni Lebe actes ocio dee 26 1 Purpose The Mid Infrared Data Analysis Software 2010 MIDAS 2010 was created for the specific use on the Mid Infrared Spectromicroscopy beam
40. mponents When the functions decay back to the baseline range scaling has some effect The slope of the synchronous component is much less and the maximum in the asynchronous component has disappeared 4 When the range scaling option is selected the maximum absolute value across the dynamic variable is compared to the Minimum Signal value If the maximum absolute value is found to be larger than the minimum signal value then the data across the dynamic variable for the particular wavenumber is divided by the difference between the maximum absolute and the minimum signal values Regular Regular scaling applies neither variance nor range scaling meth ods to the data 31 7 3 4 Axes Ticks The user has the option of determining the number of tick marks on both the X and Y axes The X and Y axes refer to the temperature or time and wavenumber axes respectively from the spectra plot Thus when the synchronous asynchronous and phase plots are viewed only the Y axis ticks input is used since these plots have wavnumbers plotted along two axes Note that the input values must be integers larger than two 7 3 5 Reverse Y Some users find it convenient to view the wavenumbers ascend on the Y axis in Plot 2 while others prefer to view them descend Selecting the Re verse Y option graphs the magnitude of the wavenumbers in a decreasing order Deselecting this option simply reflects the graph about the Z axis 8 Filtering Interface
41. nal correlation analysis ELSEVIER Chemometrics and Intelligent Laboratory Systems 30 October 1999 pages 151 152 162 163 167 168 Kovesi Peter pk cs uwa edu au bandpassfilter Department of Computer Science and Software Engineering University of Western Australia http www owlnet rice edu elec301 Projects01 image_filt matlab html October 1999 Kovesi Peter pk cs uwa edu au highboostfilter Department of Computer Science and Software Engineering University of West 77 7 8 9 E ern Australia http www csse uwa edu au pk Research MatlabEns November 2001 Kovesi Peter pk cs uwa edu au highpassfilter Department of Computer Science and Software Engineering University of Western Australia http www owlnet rice edu elec301 Projects01 image_filt matlab html October 1999 Kovesi Peter pk cs uwa edu au lowpassfilter Department of Computer Science and Software Engineering University of Western Australia http www owlnet rice edu elec301 Projects01 image_filt matlab html October 1999 Modified by Gaddi Rob gaddif rice edu ELEC 301 Rice University Decem ber 2001 Streit Katie kstreit rice edu Elter ELEC 30 Rice Univer sity http www owlnet rice edu elec301 Projects01 image_filt matlab html December 2001 78
42. on stant colours between ioslines Viewing descriptions are from Matlab Help 95 e Waterfall Similar to mesh but with a waterfall effect These are identical to those described in Section 7 3 2 8 4 Filtering Buttons When a filter is designed by manipulating the filter parameters the fol lowing buttons may be used to confirm or cancel the filter design View this Filter When a filter is designed with new parameters the filter will be compiled and plotted when the user presses the View this Filter button Select this Filter When the user presses the Select this Filter button the user confirms the filter designed as per the indicated parameters Cancel Filter Design When the user presses the X on the top right hand corner of the filtering interface the user cancels the design and no filter is applied to the data Notice how the Apply Filter checkbox is not checked 8 4 1 Parameters Versus Plotting Priority Note that if any filter parameter is modified and if the user does not press the View this Filter button before pressing the Select this Filter button mean ing the view of the filter was not updated to account for the change in parameters then the confirmed filter is designed as per the indicated pa rameters not as was shown in the figure A change in parameters always trumps the view of the plotted filter 36 8 5 Lowpass Gaussian Filter When the lowpass butterowrth filter is selected the window
43. or 13 6 Frozen Computer When the computer freezes MIDAS 2010 and Matlab are no longer re sponding The programs must be quit To do this press Ctrl Alt Delete and terminate Matlab All unsaved work will be lost 13 7 Contact Information Elise Normand designed and created MIDAS 2010 Please report all prob lems and direct all questions to her via email at elise normand usask Ca 14 Acknowledgments Thanks to Ference Borondics and to Tim May at the Canadian Light Source for their support A special and heartfelt thanks is also extended to Luca Quaroni who was the Staff Scientist on the Mid Infrared Spectromicroscopy beamline at the Canadian Light Source when MIDAS was initially created This program would not have been possible without their generous help and guidance 76 References 1 2 4 5 Eramian Mark eramian cs usask ca Associate Professor De partment of Computer Science University of Saskatchewan 29 July 2008 Eramian Mark eramian cs usask ca CMPT 487 819 Frequency Domain Filtering Associate Professor Department of Computer Sci ence University of Saskatchewan October 2009 Eramian Mark eramian cs usask ca CMPT 487 819 Prepro cessing Intensity Transformations and Spatial Filtering Asso ciate Professor Department of Computer Science University of Saskatchewan October 2009 Harrington Peter de B Urbas Aaron and Tandler Peter J Tutorial Two dimensio
44. osted relative to low frequency values If boost is less than one than a low boost filter is generated 47 Plot Setup Highboost Filter Fitter Type Highboost Cutoff 05 Create an high boost filter View Mesh Order 1 Cutoff is the cutoff fregency of the fiter where 0 lt Cutoff lt 0 5 Resolution 60 Boost 05 Order is the order of the filter The higher the order the sharper X Ticks the transition Note that the order must be an integer gt 1 Y Ticks Boost is the ratio that high frequency values are boosted relative to low frequency values Note that if Boost is less than one then a low boost fiter is generated Select this Filter Fourier Domain Filter Figure 17 Filter Interface Highboost 48 9 Data Processing and Sequence In order to correctly analyze and interpret results it is important to fully understand the data processing order and sequence of applying analytical tools They are as follows 1 Remove Static Component MIDAS 2010 checks if the user selected the Remove Static Component option and if so the program deter mines which static component the user wants to remove either Steady State or Mean Spectrum The definitions of both static components are described in Section 7 2 1 MIDAS 2010 selects all the raw spectrum data originally from the input text file and subtracts the static trace column by column If the user does not select the Remove Static Com ponent
45. own in the graph is then the absolute data and not the relative change over the dynamic variable 23 FilterGUI 2010 Plot Setup Fitter Type Lowpass Gaussian v Lowpass Gaussian Filter E Create a rotationally symmetric lowpass Gaussian fiter with ae standard deviation sigma positive Fourier Domain Filter Space Space Figure 7 Filter Interface Lowpass Gaussian 7 2 2 Apply Pre and Post Filter Selecting the Apply Pre Filter or Apply Post Filter option will bring up the filtering interface shown in Figure 7 The operations of the filtering win dow are explained in Section 8 below When the Apply Filter option is selected the current or last filter applied to the data is shown on the inter face 24 7 2 3 Average There are two types of averaging available in MIDAS 2010 and both pro duce drastically different results They are the kernel and dynamic aver ages Kernel Average Selecting kernel average smoothes the data along the dynamic variable using an averaging kernel The kernel behaves as a slid ing window where it is centered over each cell in the data matrix and the sum of products of the kernel coefficients and the underlying cells is computed 3 The size of the data is extended by mirror reflecting the data across its border at the edges of the data matrix If Average is set to some number e then the kernel has dimensions 1 e Note that e must be an od
46. ows the user to change the title of the active plot see Section 11 1 1 for the definition of the active plot The New Title window shown in Figure 25 will appear to assist the user change the plot title Since Plots A and B have the option of plotting four different types of plots namely the synchronous asynchronous phase and spectra graphs the plot title will not remain each time the plot is refreshed to a different type of plot Plots 1 and 2 are cross sections of the spectra plots Thus when a new title is given to these plots it will remain every time the plot is refreshed 73 12 5 Change Labels and Units Selecting the Change Labels menu item allows the user to change the axes labels and units The Labels and Units window shown in Figure 26 will ap pear to assist the user change the labels and units Plots A B 1 and 2 will automatically refresh in that order with the new labels and units Notice that the panel headings for Plots 1 and 2 are also changed to account for the new labels and units Changing the labels and units with this interface overrides the initial labels and units specified with the loading window 13 Dealing with Problems There are a few options available to the user when a problem is encoun tered using MIDAS 2010 They are discussed in the following sections 13 1 fspecial Error Messages If you get an fspecial error message you do not have the Image Processing Toolbox installed on your computer T
47. pem gt Start Y 14002335 15 em 10993569 y Data Setup 1 Remove Static Co Mean Sp y hs E Apply Pre Fiter Design Pre Fitter Average d 1 eng lt E Apply Post Fitter 05 Design Post Filter Z Correllation Intensities Plt Setup pl Type Synchro View Mesh 15 Resolution 15 Scale Regu e 72 unge 0 0001 X Ticks 6 Y Ticks 6 al i 1400 23 1340 44 1280 65 1220 86 116108 110129 X Wavenumber cm 1 Figure 23 Red and Blue Colour Map Nur e Time vs Absorbance Unite y lado y tan LIENS TY Se Times KC Dos 19294 122885 ms Heemer cr chter DEES he m mme y LU ev is aate ve Absorbance Unis Di d uf TELS ee ee 193 har he em N Title Synchronous Make Changes Figure 25 New Title Window 63 Axes correspond to Spectra plots Axes Labels Axes Units axis Time X axis s Y axis Wavenumber Y axis cm 1 Z axis Absorbance Units Z axis Make Changes Cancel Figure 26 Labels and Units Window to assist the user change the labels and units Plots A B 1 and 2 will automatically refresh in that order with the new labels and units Notice that the panel headings for Plots 1 and 2 are also changed to account for the new labels and units Changing the labels and units with this interface overrides the initial
48. range of the variables like wavenumbers between StartY and EndY inclusively The column vector on the Time worksheet shows the selected range of the dynamic variable like temperature or time between StartX and EndX inclusively 70 12 2 5 Synchronous Asynchronous and Phase Worksheets The content on the Synchronous worksheet shows the synchronous matrix following Step 8 in Section 9 It contains the real part of the correlation matrix The content on the Asynchronous worksheet shows the asynchronous matrix following Step 8 in Section 9 It contains the imaginary part of the correlation matrix The content on the Phase worksheet shows the phase matrix following Step 8 in Section 9 It contains the phase between the real and imaginary parts of the correlation matrix 12 2 6 Static Component Worksheet If the user removed a static component at the time the export was issued the Static Component worksheet contains the static column vector either the steady state or mean spectrum vector 12 2 7 Pre and Post Filter Worksheets If the user applied a filter to the data at the time the export was issued the contents on the Prefilter and Postfilter worksheets show the pre and post filter applied to the data respectively Note that the DC component is is the corners of the filter and not in the center of the matrix Typically Matlab shifts the filter so the DC component is placed in the center of the 71 d d Blanket
49. s respectively 5 3 Points Setup In order for MIDAS 2010 to display the correct results it is necessary to set Delta appropriately Delta is the constant A temperature or time incre ment between the columns of the data matrix A point is recorded every 6 degrees or seconds It is assumed that A is constant for the data set Since 0 represents an increment it must be positive and larger than zero and it must have units of degrees or seconds 6 Main Interface MIDAS main window is shown in Figure 5 Notice the four plotting areas named Plot A Plot B Plot 1 and Plot 2 Each plotting region was designed to help the user view the data in different ways They are explained in the following sections 17 File Insert Oe 098 DE om md r Select Data T n gt e Traces PLOT Figure Apply Selection to All stat x y onge 1 EndX 2085 e 1 DS Start Y 14002335 y f End Y 1099 3569 w oe 0 8 Data Setup R Remove Static Co 07 06 Plot 1 06 PlotA 04 0 5 1 kernel SN Apply Post Fitter 0 2 T Design Post Filter 0 3 Plat Setup 0 Type Synchro 0 0 3 02 103 04 05 06 07T 05 09 1 02r View Mesh Y Time vs Absorbance Units Resolution 6 Select Data Data Setup Plot Setup 01 22 T
50. sh Resolution 60 X Ticks 6 Select this Fiter Fourier Domain Filter Figure 13 Filter Interface Highpass Gaussian techniques described in Section 8 5 However the highpass filter is defined as a function of the lowpass filter Rhighpass x y 1 lianes y Sigma The variable sigma is the standard deviation c 41 8 9 Highpass Butterworth Filter When the highpass butterworth filter is selected the window in Figure 14 is displayed The highpass butterworth filter 7 uses the lowpass butter worth filter techniques described in Section 8 6 However the highpass filter is defined as a function of the lowpass filter Rhighpass z y 1 A C y Cutoff The cutoff variable is the cutoff frequency of the filter and must lie within the range 0 to 0 5 Order The variable order is the order of the filter The higher the order the sharper the transition The order must be an integer greater than 1 8 10 Highpass Elliptical Filter When the highpass elliptical filter is selected the window in Figure 15 is displayed The highpass elliptical filter 9 uses the lowpass elliptical filter techniques described in Section 8 7 However the highpass filter is defined as a function of the lowpass filter Rhighpass E y 1 ss y CutoffM and Cuttoffm The variables cutoffM and cutoffm are the cutoff frequencies on the major and minor axes respectively and lie within the 42 Plot Setup Highp
51. sus Plotting Priority 36 8 5 Lowpass Gaussian Filter i vu core 4C WO ae OE C953 37 86 Lowpass Butterworth Filter 37 8 7 lowpass Elliptical Filter 0 22e eoo ute ating sabes awe snk 39 8 7 1 Designing Gaussian and Butterworth Filters Using the Elliptical Filter suce ocr ers mtm ew 40 8 8 Highpass Gaussian leg 0 A AA 89 Highpass Butterworth Filler uz gos 8 10 Highpass Elliptical Filter xw RR RR 8 11 Bandpass Filter ara aaa eg 8 11 1 Alternative Designs Using the Bandpass Filter 8 12 Highboost Filter s uos ioRuR E 9 9 Red 9 RR RUBUS 9 Data Processing and Sequence 10 Plotting Buttons TUsL JBECOT BUEEOB x rh ERI Sao an e seed o 10 2 Figure Button A RR a uc a A AC RUD RN A 10 3 Apply Selection to All Button 10 4 Compare Traces Button 5 ara o die ee e 104b Property Editors tn sta des ue Sa ea 10 4 2 SAVE Button 1 moie e AS ur PEOR RR A 104 3 Export Toolbar 15 72 ots Ae Song eg 3o le ae e 11 Toolbar TIL General Teole AA E A AAA 11 1 1 Tools For the Active Plot 11 2 Colour Maps ina e da n ood ang d eod 3 ee rece ILS chase Mile caba aaa RARA a Stow 11 4 Change Labels and Units 4 44 vem mee s 49 53 53 53 53 54 56 57 57 11 5 Help and User s Eileen AAA 64 12 Menu Items 64 A AA EE he ale as 65 12 11 New and Op n sii om bor ee hene ee ed a 65 12 12 Saving Current Work Sessions 66 12 1 9 Exporting Dita s smes s
52. tallations Installing MIDAS 2010 is very easy Simply save the MIDAS2010 folder in an appropriate location This folder must be accessed by Matlab s Current Directory each time MIDAS 2010 is used The following Figure and M files should be within the MIDAS2010 folder applyFilter m ExportPassfile m averageDynamic_Final m exportPlot12_TXT m averageDynamic m exportPlot12_XLS m averageKernel m exportPlotAB_TXT m bandpassfilter m exportPlotAB_XLS m dftfilt m fastbilateral m dynamicTracesExist m fft2dcorr m EFilter m fft2dcorrelation m exportFigure12 TXT m Figure1GUI_2010 m exportFigurel2 XLS m Figure2GUI_2010 m FilterGUI 2010 m findRandomColour m gaussianfilter m getColourMap m highboostfilter m highEFilter m highpassfilter m loadfile m LoadGUI_2010 m lowpassfilter m MainGUI_2010 m MIDAS2010 m NewLabels m NewTitle m OpenGUI m OpenPlotGUI m OpenWorkspace m opus_read m paddedsize m plot2d_function m plotspectra_function m plotspectra_timeselected_compare m plotspectra_timeselected m plotspectra_wavselected_compare m plotspectra_wavselected m removeStatic m saveButton m saveFigure12 m saveFigureAB m savePic m SaveWorkspace m timestep m updateFigure1 m updateFigure2 m FigurelGUI 2010 fig Figure2GUI 2010 fig FilterGUI_2010 fig LoadGUI_2010 fig MainGUI_2010 fig MIDAS2010 fig NewLabels fig NewTitle fig MIDAS2010_UserManual pdf An example text file named RodCellData txt is also included within the MIDAS2010 folder
53. ter is applied to the data following the Compute Average step Recall from Section 8 that the filter is applied to the data in the Fourier domain The filter application steps are described in detail in Section 8 2 Timestep MIDAS 2010 re computes the temperature or time incre ment between successive points in the data matrix using A which was defined in the loading window and in Section 5 3 If the user se lected Dynamic Average with e gt 1 in the Compute Average step then the resulting data matrix is compressed along the dynamic tempera ture or time variable the increment between the columns of the data matrix is no longer Rather the increment between the columns of 50 the compressed data matrix is now 6 e Choose Selected Data Ranges MIDAS 2010 now takes the desired ranges into account The program crops the data matrix resulting from the Apply Post Filter step twice once along the dynamic vari able columns and another along the wavenumber rows That is the data between and including Start X and End X and Start Y and End Y is selected and the remaining data is ignored Special Scales MIDAS 2010 checks if the user selected the Scale op tion and if so the program determines which scale type the user wants to apply either Regular Variance or Range The scaling techniquel4 is applied to the data following the Choose Selected Data Ranges step The averaging techniques are described in detail
54. th the loading window 7 3 2 View For Plots A and B there are six ways to view the desired graphs e Mesh Wireframe parametric surface graded with colour e Contour Two dimensional isopleth or topographic map with con tour levels graded with colour e Mesh and Contour Wireframe parametric surface with a contour plot beneath the mesh both are graded with colour e 3D Contour Three dimensional contour plot graded with colour e Contour Function Filled two dimensional contour plot using con stant colours between ioslines Viewing descriptions are from Matlab Help 28 e Waterfall Similar to mesh but with a waterfall effect Figures 9 and 10 show the mesh and contour plots of two identical syn chronous and asynchronous data sets Note Resolution is only used for contour and contour related plots It determines the number of levels or isolines 7 3 3 Scale and Minimum Signal There are three scale options e Regular e Variance e Range Both the variance and range scale options depend on the Minimum Scale input Variance The correlation intensities between two large peaks that are only slightly correlated may be larger than the correlated intensity be tween two smaller peaks that are highly correlated The variance scaling on the correlation intensities is evaluated for the univariate case meaning row by row in the data matrix Scaling the data by the Euclidean length as in the square root of the
55. tools are listed on the main interface DA 10 Plotting Buttons 10 1 PLOT Button Pressing the PLOT button will plot or refresh the respective plot while ap plying the current analytical and viewing settings The data is processed according to the steps outlined in Section 9 10 2 Figure Button When the Figure button is pressed the data is re processed as outlined in the steps in Section 9 However rather than being plotted in the appro priate location on MIDAS main interface the data is plotted in a separate figure which comes to view Separating the plot from MIDAS main in terface is advantageous it grants the user access to Matlab s regular tools Matlab treats this new figure as any other figure That is to say while the plot remains the active figure the user can enter commands in Matlab s Command Window to further alter the plot Additionally since the plot is isolated from MIDAS main interface it can be saved exported printed etcetera as desired 10 3 Apply Selection to All Button When the Apply Selection to All button is pressed MIDAS 2010 copies the ranges selected for Plot A with Start X End X Start Y and End Y and 53 SSC Fare l Sal Synchronous i d 45t 1 a E SS em 8 1 E Ti bh X Time s 7 Time vs ot irits CM Res 15 Select Data a Set 2 ege Regar ER Ae Seal a z 13 in 0 a o P en Ee wel 0 000 Ji a ficks 6 licks amp nc S C De Fil X Ticks
56. trace across the dynamic variable When the input data is loaded from the text file each column is treated as a vector and the average value is returned for 22 each vector This forms a mean trace along the dynamic variable That is for each 6 temperature or time increment the average over all wavenum bers is computed and together these averages form a mean trace along the temperature or time variable Removing the mean spectrum from the data effectively subtracts the mean value from the respective column Therefore the resulting data represents the change relative to the mean trace rather than the absolute data Keeping the mean spectrum leaves the data unaltered In this case the data shown in the graph is then the absolute data and not the relative change from the mean trace Steady State The steady state component is defined as the very first col umn of the input data matrix which corresponds to the initial temperature or zero time It is essentially the DC component Removing the steady state component effectively subtracts the first column from each column of the data matrix Therefore the resulting data represents the change relative to the DC component rather than the absolute data Removing the steady state component automatically sets the first column of data to zero since it cancels out itself through the subtraction Keeping the steady state component means the DC component is left in the data In this case the data sh
57. utoff frequencies on the major and minor axes respectively and lie within the 39 range 0 lt cutoff lt 1 Order Order is the order of the filter The higher the order the sharper the transition The order must be an integer greater than 1 Alpha The variable alpha is an angle in radians which rotates the filter counter clockwise through alpha Obviously and angle of zero does not rotate the filter Xoffset and Yoffset The Xoffset and Yoffset variables offset the filter in the x and y directions respectively An offset of O corresponds to the center of the filter and an offset of 1 corresponds to the edge of the filter A positive offset shifts the filter in the positive direction while a negative offset shifts the filter in the negative direction 8 71 Designing Gaussian and Butterworth Filters Using the Elliptical Filter A Gaussian filter can easily be designed with the Elliptical filter if cutoffM cutoffm 0 5 order 1 and alpha Xoffset Yoffset 0 Similarly a butterworth filter can be designed with the same parameters but with order gt 1 8 8 Highpass Gaussian Filter When the highpass Gaussian filter is selected the window in Figure 13 is displayed The highpass Gaussian filter uses the lowpass Gaussian filter 40 4 Highpass Gaussian Fiter Filter Type Highpass Gaussian Sigma 0 Create a rotationally symmetric highpass Gaussian filter with standard deviation sigma positive View Me
58. utterworth 38 Plot zue Lowpass Elliptical Filter Fitter Type Lowpass Elliptical Coti 025 Create an lowpass elliptical fiter View Contour x Cutoffm CutoffM is the cutoff fregency on the major axis where Resolution 60 Order Kee XTicks 6 Cutoffm is the cutoff egenen on the minor axis where b Alpha 2 0 Cutoffm lt 1 Saa 8 Xoffset Order is the order of the fitter The higher the order the sharper Yottset the transition Note that the order must be an integer gt 1 Alpha is an angle in radians which rotates the filter counter clockwise through alpha Xoffset offsets the filter in the x direction An offset of 0 corresponds to the center and an offset of 1 corresponds to the edge of the filter A positive offset shifts the filter in the positive direction Yoffset offsets the filter in the y direction An offset of O corresponds to the center and an offset of 1 corresponds to the edge of the filter A positive offset shifts the filter in the positive direction Fourier Domain Filter T L 1 1 1 fi L 200 250 300 350 400 450 500 Space Figure 12 Filter Interface Lowpass Elliptical 8 7 Lowpass Elliptical Filter When the lowpass elliptical filter is selected the window in Figure 12 is displayed The user can design an nth order elliptical lowpass digital but terworth filter 9 using the following variables CutoffM and Cuttoffm The variables cutoffM and cutoffm are the c

MIDAS 2010 User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents