DISCRIM: A Matlab Program for Testing Image Discrimination Models
Contents
1. and a a minimum and maximum output contrast e g due 0 to a veiling illumination on the display Thus the output of the nonlinearity is oa r C ma 7 ore e E a x is the input level a number between 0 and 1 xX is the liftoff level y is the degree of nonlinearity 2 3 3 is a typical range of values for a CRT system and such values are often built into devices such as DLP projectors which Discrim User s Manual Page 9 don t have an implicit pixel nonlinearity C and C are the minimum and maximum output contrast values y is the resulting output contrast a number between 1 and 1 The resulting gamma curve is plotted in the lower half of the window Output Noise The next and final stage of the display i model is output noise The model allows for Gaussian noise to be added to the displayed pixels after the point goan C No output noise nonlinearity has been applied This may be used to model RMS noise contrasti 0 1 imperfections in the display device but can be used for other E modeling applications as well e g models of medical imaging devices The user specifies the root mean square RMS noise contrast 1 e the standard deviation of the noise in units of image contrast Menu Display Control The Display Control menu has three sets of items The first Display control three items relate to how the image contrast is scaled prior to display w 4s1s on the main window The2. M A de Vries D J amp Soares E J 1997 Comparison of the channelized Hotelling and human observers for lesion detection in hepatic SPECT imaging Proc SPIE Image Perc 3036 14 20 Landy M S 2003 lt A tool for determining image discriminability http www cns nyu edu msl discrim discrimpaper pdf Myers K J Barrett H H Borgstrom M C Patton D D amp Seeley G W 1985 Effect of noise correlation on detectability of disk signals in medical imaging Journal of the Optical Society of America A 2 1752 1759 Rohaly A M Ahumada A J Jr amp Watson A B 1997 Object detection in natural backgrounds predicted by discrimination performance and models Vision Research 37 3225 3235 Rolland J P amp Barrett H H 1992 Effect of random inhomogeneity on observer detection performance Journal of the Optical Society of America A 9 649 658 Wagner R F amp Brown D G 1985 Unified SNR analysis of medical imaging systems Phys Med Biol 30 489 518 Wagner R F amp Weaver K E 1972 An assortment of image quality indices for radiographic film screen combinations can they be resolved Appl of Opt Instr in Medicine Proc SPIE 35 83 94 Wyszecki G amp Stiles W S 1982 Color Science Concepts and Methods Quantitative Data and Formulae Oo Ed New York Wiley
3. This option allows the user to threshold an image clipping values that are too low and or too high Image Number 3 128x128 low contrast airport Clip low values Threshold Contrast Level 1 l Clip high values Threshold Contrast Level 1 Okay Cancel iix Undo This option allows the user to undo the most recent image edit reverting to the previous contents of that edited image Image 3 low contrast airport Okay Cancel Operation Add Grating Menu Display Characteristics The display characteristics menu allows the user to specify Deeley characteristics C viewing Geometry parameters that relate to the sensor and display model There are five F options each of which invokes a pop up window The first four options a correspond to the four stages of the display model and the fifth describes the viewing geometry Viewing Geometry The viewing geometry window specifies the viewing distance to the simulated CRT the pixel sampling and the image size The screen related variables may be specified in units of visual angle deg distance cm or pixels Since these quantities are all interrelated changing any one variable is likely to change others And which others are changed depends on which variables one chooses to hold constant In most cases the viewing distance and the number of image pixels are held constant But it is a good idea to check all the variables once you ve made changes to ensure that y
4. 128x128 airport 2 directory window The directory window includes c 3 128x128 low contrast buttons that allow the user to choose the images Okay Cancel displayed in the two display areas on the main window These choices take place when the user hits the Okay button Discrim User s Manual Page 5 Menu Edit Image The edit image menu allows the user to modify the contents of the tines Ps Clear currently defined images or to create new ones There are eight options in Add Grating this menu With any of the Add or Combine options if an image number is numer Combine File specified that is new it will be created If no image is specified one will be combine Image Scale created with a new number Clip Clear This option clears the current image setting all pixel values Unda to zero Add Grating This option adds a grating to an image The grating may either be a sine wave a square wave or a triangle wave The user may specify all of the 7 en 1 oyolesideoree relevant parameters of the grating accel iix Image Number 3 128x128 low contrast airport C eycles image width Orientation deg o 0 pi gt f 80 Ce deg lee lo 0 360 o image center Type amp sine C square triangle Peak Contrast Ea 0 E 1 Okay Cancel Add Gabor This option adds a Gabor patch a sine mmm aa Image Number 3 128x128 low contrast airport wave windowed b
5. a new image giving it a name and optionally ImageNumber 3 specifying a nonstandard image size Rows 128 Columns 128 Load This option allows the user to load TE in an image from disk This image will either mageNumber 2 airport2 replace a currently defined image or will be used par O O to create a new image A Browse button allows _ Browse the user to search for the appropriate image file If the filename ends in mat it is assumed to be a Dow _ Cancel _ Matlab save file that defines an array variable called discrim_img It will complain if this variable is not defined in the file For all other filenames the Matlab routine imread is used Thus images in all of the formats handled by imread may be read including BMP HDF JPEG PCX TIFF and XWD files If the image is new or the image name was blank it will be given the filename as the image name Save This option allows the user to save an image to disk The same image types may be MaseNumber 2 aipon2 ici xi saved as can be loaded see above pan Fe res Delete This option will delete the minam p current image Format Matlab BMP HDF JPEG PCX TIFF XWD Okay Cancel Directory This option opens a window aioi listing all currently defined images If this window Left Right is kept open while the user works any changes c 1 128x128 airport 1 made by other commands will be registered in this Cc c 2
6. is linearly scaled to pixel value so that a contrast of zero maps to pixel value 128 1 maps to zero and 1 maps to 256 The Main Window The main window of dis Cr j m Model Model nran Image Edit Image Display Characteristics Display Control is generally kept visible the entire Image Name Image Name time one uses this program This window contains two images The EE Z2 Ee E 256x256 256x256 images that are displayed may be altered by typing in a new image number If a new number is entered an image is created with size determined by the current viewing conditions see the description of the Viewing Geometry window under Menu Display Characteristics below The image name may be Display AT entered here as well Image Displayed Raw Image One of the two images on the Current model Single Filter Uniform Masking main window is designated the Calculate dprime 1 7366 Quit Discrim User s Manual Page 3 current image This image is the default for many of the image manipulation tools The current image is displayed with a magenta border The user can switch to the other image by clicking on that image When an image manipulation tool is used the image that is created or modified will be displayed and made into the current image unless both main window image slots are already occupied by other images The images that are displayed and the image designated as the current image may also be specified in
7. multiplied Note that if the image file is larger than the image or if the spatial offset shifts part of the image file off of the image those pixels lying outside the image are discarded Discrim User s Manual Page 6 151 x Combine Image This option combines one currently defined image the From image with another currently defined image or creates a new image called the To image The user may specify a spatial offset to shift the From image by default that oe image 1s centered The pixels from the From image may Overlay with Transparency To Image Number 3 256x256 low contrast airport From Image Number 2 256x256 airport 2 Spatial Offset Rows 0 Columns 0 Action Add be added to the To image pixels overlaid onto the To nen Okay Cancel image pixels replacing or occluding them overlaid with transparency so that From image pixels with a value of zero do not replace To image pixels or multiplied Note that if the From image is larger than the To image or if the spatial offset shifts part of the From image off of the To image those pixels lying outside the To image are discarded Scale This option allows the user to linearly eee scale a currently defined image first multiplying by a MaseNumber s 128x128 low contrast airpon contrast scale factor and then adding a contrast shift SonrastScae s amount Contrast Shift o Okay Cancel iix Clip
8. DISCRIM A Matlab Program for Testing Image Discrimination Models User s Manual Michael S Landy Department of Psychology amp Center for Neural Science New York University Discrim 1s a tool for applying image discrimination models to pairs of images to predict an observer s ability to discriminate the two images It includes the ability to model both the human observer as well as the sensor and display system that is used to detect and display the image materials e g night vision equipment A general introduction to the design of discrim is described by Landy 2003 It is our hope that display modelers and evaluators will be able to use discrim to test the quality of displays and their usefulness for particular visual tasks We are making the code and documentation freely available to the general public at http www cns nyu edu msl discrim We hope that others will make use of the software and will let us know what other capabilities would be useful Discrim implements a simple sensor and display model consisting of four stages These are in order of application 1 a linear spatial filter 2 Poisson input noise 3 a point nonlinearity and 4 Gaussian output noise At present discrim includes a single image discrimination model called the single filter uniform masking SFUM model Ahumada 1996 Ahumada amp Beard 1996 1997a b Rohaly Ahumada amp Watson 1997 see http vision arc nasa gov personnel
9. Image Edit Image parameters of the discrimination models There is one item Single Filter Unif Masking for each available discrimination model The Single Filter Model with Uniform Masking has Miimii seven parameters This model uses a single channel with a Center spread min contrast sensitivity function CSF modeled as a difference Frequency cut cpa TA of Gaussians as suggested by Barten The CSF parameters include the center spread in min arc and its corresponding high frequency cutoff the spread ratio the Amplitude ratio 0 685 ratio between the two Gaussian sigma parameters and the amplitude ratio of the two Gaussians The contrast sensitivity specifies the sensitivity at the peak of the CSF After filtering by the CSF the pixel differences between Beta the two images are summed with a power of beta The Okay resulting value is then scaled down by a masking factor Ix Spread ratio Contrast sensitivity maximum 4 Mask contrast threshold 0 on mol 2 z nase na n 2 Discrim User s Manual Page 4 determined by the degree to which the standard deviation of the contrast in the left hand image exceeds a mask contrast threshold Menu Image The image menu allows the user to manage the current set of Image Edit Image defined images There are five options in this menu New Load Save Delete Directory New This option allows the user to define mi x
10. al code filtmod1 htm The model includes a contrast sensitivity function as well as a simple model of pattern masking SFUM was designed to estimate the value of d for discriminating two given fixed images For example in evaluating a lossy image compression scheme SFUM will provide an estimate of the ability of an observer to discriminate an original image from its compressed distorted counterpart The resulting d pronounced d prime value indicates the degree of discriminability A d value of zero indicates the two images are completely indiscriminable so that an observer would be 50 correct 1 e guessing on a two alternative forced choice task d values of 1 and 2 correspond to performance of 76 and 92 correct respectively T Address correspondence to Michael S Landy Department of Psychology New York University 6 Washington Place Rm 961 New York NY 10003 landy nyu edu 212 998 7857 fax 212 995 4349 Discrim User s Manual Page 2 Discrim 1s written in Matlab and is invoked assuming your Matlab path is set appropriately by typing discrim at the Matlab prompt Discrim has the ability to read and save images and to do some rudimentary image processing to create the images to be discriminated Discrim assumes that images are presented at a reasonably large viewing distance on a CRT or other rectangularly sampled image display device Image display geometry may thus be specified in units of di
11. ay window and will not be visible When the user hits the Calculate button the two displayed images are compared using the current image discrimination model However if an image is smaller than the current Viewing Geometry image size it will be extended with zeros If an image 1s too large the pixels that lie outside the display window will not be used in the discriminability calculation The SFUM Model Implementation The SFUM model was designed to estimate the discriminability of a pair of fixed images The display model involves two possible sources of noise input Poisson noise and output Gaussian noise When either or both of those noise sources are enabled the intent of the discrim program is to allow the user to estimate the discriminability of the two input images under conditions of stochastic variability due to the noise source s and other image distortions That is on any given trial an observer will see a different retinal image due to the variability of the noise from trial to trial If we simply added different independent Poisson and or Gaussian noise samples to each of the two images and then applied the SFUM model SFUM would attempt to estimate discriminability not only of the underlying images but of the two noise samples as well Clearly this is not appropriate What is of interest is the observer s ability to discriminate the underlying scenes despite the noise not their ability to discriminate the noise samples T
12. et al 2003 King de Vries amp Soares 1997 Myers et al 1985 Rolland amp Barrett 1992 Wagner amp Weaver 1972 for reviews see Eckstein Abbey amp Bochud 2000 Wagner amp Brown 1985 These models provide an estimate of d given the input images and descriptions of the noise variance spatial correlation etc Thus for these models discrim 1s already set up to provide the required information and the issue of using identical noise samples for the two input images shouldn t arise It would be a relatively simple task to add such models to the discrim model palette References Ahumada A J Jr 1996 Simplified vision models for image quality assessment In J Morreale Ed SID International Symposium Digest of Technical Papers 27 397 400 Santa Ana CA Society for Information Display Ahumada A J Jr amp Beard B L 1996 Object detection in a noisy scene In B E Rogowitz amp J Allebach Eds Human Vision Visual Processing and Digital Display VII 2657 190 199 Bellingham WA SPIE Ahumada A J Jr amp Beard B L 1997a Image discrimination models predict detection in fixed but not random noise Journal of the Optical Society of America A 14 2471 2476 Ahumada A J Jr amp Beard B L 1997b Image discrimination models Detection in fixed and random noise In B E Rogowitz amp T N Pappas Eds Human Vision Visual Processing and Digital Display VII 3016 34 43 Bel
13. hus the SFUM model is not well suited to the problem at hand However we have implemented the SFUM model in a way that should allow it to provide reasonable estimates We do this by using the same sample of noise Poisson and or Gaussian for both images Thus differences between the two noise samples are not there to inflate the d estimates Gaussian noise is an additive process that is independent of the image content It is a simple matter to generate a Gaussian noise image and add it to both input images On the other hand Poisson noise depends on the image content The variance of the noise added to any given pixel is equal to the value of that pixel This means that the use of the same noise image for both input images is not an accurate reflection of Poisson statistics We have settled on an approximation that we feel is adequate for the sorts of threshold detection tasks for which discrim is most appropriate When Poisson noise is used with the SFUM model the two input images are first blurred using the current MTF Then the image in the left hand window is subjected to Poisson noise The difference between the noisy image and the blurred left hand image the error image 1s treated as Discrim User s Manual Page 11 an additive noise source That error image is then added to the individual blurred images to simulate a Poisson noise source that perturbs both images identically The reason the left hand image is used to generate the Poiss
14. lingham WA SPIE Discrim User s Manual Page 12 Barrett H H Yao J Rolland J P amp Myers K J 1993 Model observers for assessment of image quality Proceedings of the National Academy of Sciences USA 90 9758 9765 Bochud F O Abbey C A amp Eckstein M P 2000 Visual signal detection in structured backgrounds III Calculation of figures of merit for model observers in non stationary backgrounds Journal of the Optical Society of America A 17 193 205 Burgess A E 1999 Visual signal detection with two component noise low pass spectrum effect Journal of the Optical Society of America A 16 694 704 Burgess A E Li X amp Abbey C K 1997 Visual signal detectability with two noise components anomalous masking effects Journal of the Optical Society of America A 14 2420 2442 Eckstein M P Abbey C K amp Bochud F O 2000 A practical guide to model observers for visual detection in synthetic and natural noisy images In J Beutel H L Kundel amp R L van Metter Eds Handbook of Medical Imaging Vol I Physics and Psychophysics pp 593 628 Bellingham WA SPIE Press Eckstein M P Bartroff J L Abbey C K Whiting J S Bochud F O 2003 Automated computer evaluation and optimization of image compression of x ray coronary angiograms for signal known exactly tasks Optics Express 11 460 475 http www opticsexpress org abstract cfm URI OPEX 11 5 460 King
15. ltiplied by one applied to the vertical frequencies For the DoG filter each of the constituent Gaussian filters 1s Cartesian separable For the Gaussian and DoG filters the user may specify whether the horizontal x and vertical y filters are identical in which case the parameters of only the horizontal filter are specified or not The Gaussians are specified by the standard deviation sigma in either the spatial or Spatial frequency domain in a variety Of ppp units The DoG filter consists of a Gaussian Flat x MTF y MTF with a high cut off frequency minus a ee second Gaussian with a lower cut off cassin sim rs 75 eycesideges degrees frequency The user specifies the relative D a eyeleslem R amplitude of the two filters the peak ratio peaknaioi fos eyolesimagewidh C image widths a value which must lie between zero and Path A one The DoG filter is scaled to have a peak P kenam OOo value of one For the empirical filter the user specifies the location of a text file that describes the filter The format of the file is simple Each line contains either two or three values The first is the spatial frequency in cycles deg The second is the horizontal MTF value which must lie between zero and one If all entries in the file have only two values then the program assumes the vertical and horizontal filters are identical Otherwise all lines should contain three values and the third value is the ve
16. minant Integration time 4 msec wavelength of the light the quantum efficiency of EE so nm the individual pixels 1 e the percent of the photons incident on the square area associated Lan 70 g with the pixel that are caught and the sensor Pupil diameter e mm aperture diameter The calculation based on the formula in Wyszecki and Stiles 1982 only makes sense if the image consists primarily of visible light Thus for infrared sensitive equipment dominant wavelengths above 700 nm the user must supply the mean quantum catch The input images are specified in terms of image contrast Input image pixels are linearly mapped to expected quantum catches so that a contrast of 1 corresponds to an expected catch of zero and a contrast of zero corresponds to the mean expected quantum catch Gamma The user may specify the memmen ioii nonlinearity applied to individual pixels Typically both image sensors film vidicons etc and image displays CRTs in particular are characterized by a c Disabled Enabled so called gamma curve Discrim includes a oe o Ea Bl single nonlinearity in its sensor and display model Gamma 3 of S One can think of it as a lumped model that Min Contrast 8 4 1 combines both the sensor and display nonlinearities az Max Contrast Es 1 d 1 We implement a generalization of the gamma curve by allowing for an input level below which no output occurs which we call liftoff
17. new samples of Poisson and or Gaussian noise are computed changing the displayed images if the appropriate image display selection has been specified Working with Images of Different Sizes Discrim has the ability to work with images varying in size The set of images in the current directory can include images with different numbers of rows and or columns Throughout it is assumed that the pixel sampling is identical So for example when images are added the pixels are added one at a time in order Images are also assumed to Discrim User s Manual Page 10 be centered on the display and with respect to each other although spatial offsets are included with the Add Gabor Add File and Add Image commands For display and image discriminability calculations however discrim has a fixed image size as specified in the Viewing Geometry window The larger of the two image display dimensions the larger of the number of horizontal and vertical pixels specified in the Viewing Geometry controls the scaling of images in the display areas of the main window Thus if the image size is changed in the Viewing Geometry window displayed images may become smaller or larger When images are displayed they are shown centered in the corresponding window with a purple or gray border outside of the defined image area If an image is larger than the size specified by the current Viewing Conditions pixels beyond that size will fall outside the image displ
18. next five items are used to select what ote form of the images are displayed And the last item allows the user l to display a new set of noise samples eee By default images are displayed As Is meaning that an after Poisson noise image contrast of 1 is displayed as black and 1 as white The After gamma curve Display Control menu allows the user to either stretch the contrast 25r Saussian noise as much as possible leaving 0 as mid gray or to linearly scale eee displayed contrast so as to use the full range of displayable contrasts mapping the lowest pixel value to black and the highest to white Note that the full range setting effectively undoes the liftoff value of the gamma function at least as far as the appearance of the displayed images is concerned By default the images displayed on the main window are the input images unmodified by the various distortions in the image display model However using the Display Control menu the user may select for display the image after each stage of the image display model after the MTF has been applied after the Poisson noise has been added after the gamma curve nonlinearity or after the Gaussian output noise has been added The two images each have their own separate samples of Poisson noise since Poisson noise variance depends on image content On the other hand the same Gaussian output noise sample is used for both displayed images Finally if the user specifies New Noise
19. on noise is due to an asymmetry inherent in the SFUM model The SFUM model treats the left hand image as a mask and the difference between the two images as a signal As long as the mask or background is kept constant then d is proportional to the strength of the signal Thus to determine the strength of signal required to produce a d of 1 0 for example one need only divide the current signal strength by the current value of d Note that each time discrim calculates a d value new samples of Poisson and or Gaussian noise are used Thus the user can average over several such calculations to guard against an outlier value due to an atypical noise sample Also note that the noise samples used to calculate the SFUM model are not the same samples as are used for image display in the main window Finally note that the images are not clipped prior to applying the SFUM model so that some pixels may have contrast values above or below 1 which is of course not a displayable contrast Additional Observer Models The discrim program 1s set up so as to allow for the addition of additional vision models In particular there is a large literature mostly from the medical imaging community of visual detection and discrimination models for visual targets in patterned and noisy backgrounds Barrett Yao Rolland amp Myers 1993 Bochud Abbey amp Eckstein 2000 Burgess 1999 Burgess Li amp Abbey 1997 Eckstein
20. ou have things set the way you want them When the number of image rows or columns is changed this has consequences for other things such as how images are displayed the interpretation of the MTF and the model calculations The Viewing Conditions window Output Moize Discrim User s Manual Page 7 assumes that both images are square so that HA changes to horizontal size also change the Viewing distance em 100 vertical size and vice versa with square pixel Horizontal pixelsideg 30 Square pixels sampling so that changes to horizontal pixel Vertical pixelsideg 30 sampling density change the vertical pixel Horizontal pixelsfom 17 19 sampling density and vice versa Check boxes Vertical pixels em 710 are supplied to allow the user to turn off this romagecetimns 128 Square image behavior Number of image rows 128 i Horizontal image size cm 7 447 MTF modulation transfer function Ooo aA The input spatial filter may be used to MIMIC Horizontal image size deg EE the effects of the optics of the image sensor Vertical image size deg 4 267 The MTF window gives four choices of filter Okay type Flat no filtering Gaussian Difference of Gaussians DoG and Empirical The Gaussian and Empirical filters are Cartesian separable Very loosely speaking Cartesian separable means that the filter may be specified as a product of a filter applied to the horizontal frequencies mu
21. rtical MTF value The supplied MTF is interpolated and or extrapolated as needed using the Matlab interp spline interpolation routine The file should be sorted in ascending order of spatial frequency The MTFs are graphed in the bottom half of the MTF window The MTF is applied in the Fourier domain That is the input images clipped to the size of the display window and or extended with zeroes to that size are Fourier Discrim User s Manual Page 8 transformed The transforms are multiplied by the MTF and then inverse transformed This is performed using a discrete Fourier transform DFT and hence is subject to the usual edge effects of the DFT These edge effects can be ameliorated by using a windowing function e g use the Add Gabor with zero spatial frequency to create a Gaussian image and use Combine Image to multiply images with this Gaussian window Input Noise The user may specify or have a mee the program compute the mean quantum catch of the individual sensor pixels When that quantum catch is low as it must be in the low light conditions for which night vision equipment is Mean photon count per pixel 18 809 designed the effects of the Poisson statistics of light become important To compute the mean C No input noise C Based on mean photon count per pixel Calculated e Mean luminance 0 1 cd m 2 quantum catch the user specifies the mean luminance the integration time the do
22. splay pixels distance across the display device in cm or angular distance as viewed by the observer in deg Images and Image Numbers Discrim maintains a library of images The images in this library are raw input images That is they are meant to be descriptions of a scene before that scene is distorted by the various elements of the image sensor and display model 1 e by blur noise or nonlinearities Images may be created by reading in pre computed images from disk by combining images arithmetically and by adding simple targets to images The program keeps track of images the user has created referring to these images by their image number These image numbers are positive integers used to identify the images Users may optionally give images a name as well which 1s displayed next to the image number on each of the windows An image directory may be displayed as well listing all the images that have been defined In many cases when a tool requests an image number and the field is left blank by the user a new image is allocated instead and given the first available image number Images are kept as Matlab arrays in which each pixel value represents image contrast That is a pixel value of zero represents a mid gray pixel a value of 1 represents a black pixel and a value of 1 represents a pixel with double the luminance of the mid gray When images are imported from or exported to standard 8 bit formats e g JPEG or TIFF contrast
23. the image directory window see Menu Image below Below the images is an indication of the way the image contrast is displayed and which version of each image is displayed see Menu Display Control below as well as the current observer model see Menu Model below The Calculate button is used to calculate the perceptual difference in d units between the two displayed images The main window has six menus that we describe next Many of the menu options result in a pop up window These windows may be used to perform various image manipulations In the various image manipulation windows invoked from the mage and Edit Image menus no actions take place until the user hits the Okay button and if the action is successfully completed the window disappears The Cancel button may be used to dismiss the window without any action taking place However the items in the windows invoked by the Model Parameters and Display Characteristics menus are changed the moment they are changed on screen Menu Model The Model menu is used to choose the observer model to Model Model Parameters Image be applied to the image pair For now the only model that has _ Single Filter Unif Masking been implemented is a simple single channel SFUM model developed by Al Ahumada It would be a relatively simple matter to add other models to discrim when the time comes Menu Model Parameters The Model Parameters menu is used to alter the ModelParameters
24. y a Gaussian envelope to an image The Samot ro o come 0 envelope may be an arbitrary Gaussian The user specifies cassan Envelope its orientation and the spread parameters along and across gpeney um cep that orientation The sine wave modulator is specified asin Somate the Add Grating command This patch is normally centered eee in the image but may be shifted an integral number of on BE a pixels horizontally and vertically Note that typical image oiemaionideg o ouf ojt Mo EE C em C horizontal pixels Units cycles deg C eycles cem Phase deg relative o ll el 360 row numbering is used so a positive row offset shifts the t Gaor eenen a patch downward _ owy cmoa Combine File This option combines an image memmmmmmm ERE read from disk with a currently defined image or s Izas low contasiaipon creates a new image If a currently defined image 1s SP talottset Rows o Columns o specified the user may specify a spatial offset for the Pah ciDocuments and Setingsmir image from the disk file by default that image is Fiename fsvsoweai centered The pixels from the image file may be added acion aaa to the image pixels overlaid onto the image pixels C Overlay 7 C Overlay with Transparency replacing or occluding them overlaid with anne transparency so that image file pixels with a value of Okay Cancel _ zero do not replace image pixels or
Download Pdf Manuals
Related Search