Home

1 Video Eyetracker Toolbox

1. MONITOR TYPE The manitor that you ere currently using I your maritar Fig 9 Documentation within vsgDesktop Non Computer based Custom Display Device If you are planning to use the Toolbox with a custom presentation device you cannot calibrate the device directly Instead you must define the bounding dimensions of the device using the Toolbox s programming interface This can be achieved using the function vetSetDeviceParameters and passing the parameter CRS deUser followed by the viewing distance and finally the Width and Height of the device Setting the Toolbox s presentation device to CRS deUser activates the vetSetCallbacklnitialiseDisplay vetSetCallbackClearDisplay and vetSetCallbackDrawTarget event call back procedures during the subject calibration process These call backs can then be used to present targets at defined locations to the subject during calibration More information regarding vetSetDeviceParameters and the callback functions can be found in the Appendix A at the end of this manual 20 Video Eyetracker Toolbox Calibration Positioning and Focusing your Subject Imaging The EyeLock headrest is designed to image the eye in a way that is both safe and minimally invasive to the subject It achieves this via an infrared mirror which appears transparent to the subject while keeping the camera and illumination sources out of their field of view The Toolbox is designed for ro
2. During tracking the Toolbox is able to accommodate fast eye rotations sudden head movement and changes in pupil size By imaging in the infrared it is also relatively immune to changes in ambient illumination and can accommodate a wide range of spectacles or contact lenses The system is intelligent enough to know whether or not an eye is present in the video images This subject is therefore free to blink or take breaks between eye tracking tests if necessary without the need to recalibrate each time The system can return eye position in Fick Helmholz or screen position in millimetres and will automatically return Not Tracked if it detects that an eye is not present in the video image videoEyetrace will automatically terminate saccade smooth pursuit or bitmap image tasks at the end of their defined duration Alternatively any eye tracking task can be terminated using the stop icon on the main toolbar 9 Fig 42 Stop Icon 47 Video Eyetracker Toolbox videoEyetrace Saving and Loading videoEyetrace can save and load Data files These can be saved and opened either via the File menu or the Save and Open buttons on the toolbar Data Files ved The default videoEyetrace data files are text files containing all the results collected from an eye tracking experiment The header section summarises the experiment in terms of the type of stimulus used and the duration of the experiment The header section also states
3. error Video Source not selected end 68 Video Eyetracker Toolbox Communication o Calibrate the video eyetracker vetCreateCameraScreen errorCode vetCalibrate if errorCode 0 error Calibration not completed end Create the Mimic Screen window vetCreateMimicScreen vetSetMimicScreenDimensions 400 0 400 300 vetSetFixationPeriod 300 vetSetFixationRange 10 vetSetMimicPersistenceType CRS ptMotionAndFixations Subject Calibration VET Camera Window Gy calibrating step 16 of 20 Sy VET Camera Window i E Pupil Target Diameter 2 mmm e Save as Default Calibration Cancel Ok Fig 60 250 Hz Live Camera and Calibration Windows Calibration in Progress The next section of code defines the colours size and shape of all stimuli and draws the stimuli into video memory In this example we have chosen a green circle as the fixation target and red circles as saccade targets All circles have a viewed diameter of 0 3 degrees Set the palette and pen levels for drawing red and green dots on a mean grey background GreyLevel 1 GreyRGB 0 5 0 5 0 5 crsPaletteSetPixelLevel GreyLevel GreyRGB RedLevel 2 RedRGB 1 0 0 0 0 0 crsPaletteSetPixelLevel RedLevel RedRGB GreenLevel 3 GreenRGB 0 0 1 0 0 0 crsPaletteSetPixelLevel GreenLevel GreenRGB 69 Video Eyetracker Toolbox Communication Draw the fixation point and target on
4. Fig 56 Select Stimulus Monitor Select the appropriate monitor by clicking the relevant button The following lines of code will now display the stimulus image on the chosen stimulus monitor 61 Video Eyetracker Toolbox Communication Pick a stimulus image and get the full path to the file The image file must be on your MATLAB path ImageFile which Glasgow bmp Read the selected image file getting both the image and the palette data Image Palette imread ImageFile 7 Create a figure window to display the stimulus Load the bitmap image onto the selected monitor Scaling appropriately FigurelHandle figure 1 AxesHandle axes set AxesHandle Position 00 1 1 set FigurelHandle MenuBar none ToolBar none Position SelectedPos imagesc Image colormap Palette axis oft The same stimulus image will now be loaded onto a Mimic screen on your Windows monitor Create Mimic Screen and load Bitmap image vetCreateMimicScreen vetSetMimicScreenDimensions 400 0 400 300 vetSetFixationPeriod 300 vetSetFixationRange 10 vetSetMimicPersistenceType CRS ptMotionAndFixations vetLoadBmpFileToMimicScreen ImageFile 1 Notice that the Mimic screen has been set to show any fixations which last for at least 300 milliseconds within a 10mm cluster The Toolbox will now track for 30 seconds All eye movements and fixations will be output to the Mimic wind
5. e The draw mode will be set to CRS CENTREXY 33 Video Eyetracker Toolbox Calibration e Any page cycling entries and stored LUTs should be considered lost Dual VGA Video Eyetracker Toolbox creates a separate window to display the calibration targets This window is then destroyed after subject calibration has been completed Consequently it is not necessary to reset the stimulus window after calibration Custom Device The Toolbox has no direct influence over the state of custom devices that are used with it Cleaning up of the display after Calibration is left entirely to the responsibility of the user 34 Video Eyetracker Toolbox videoEyetrace d In this section Overview 36 The videoEyetrace Window 36 Initialising the Video Eyetracker Toolbox Camera 37 Changing Stimuli and Settings 40 Changing View Options 41 Graph Tab 42 Configuration Tab 43 Mimic Tab 43 EIB Tab High Speed System Only 44 Eye Tracking 45 Saving and Loading 48 Data Files ved 48 Custom Results Data Files ved 49 MATLAB Data File mat 49 Mimic Image bmp 49 35 Video Eyetracker Toolbox videoEyetrace Overview It is important to remember that the software component of the Video Eyetracker Toolbox is implemented to enable eye tracking functionality to be accessed and controlled from MATLAB or any Windows programming tool that supports Microsoft COM However for completeness and ease of use the Toolbox is also supp
6. the Library Topics section on Triggering in the Documentation supplied in your vsgDesktop for further details The combined values for these three lines are returned to your program as the DigitallO parameter in the data set for each frame Fig 47 displays the possible values for this parameter and the corresponding states of the three channels DigitallO Channel State Value Line 2 Line 1 Line 0 o0 fF WM o E O O OO ES o Fee NC BN Foe Ec NOS Fig 47 Combined Digital I O Values The fourth digital I O line is an output trigger This line is set high whilst the Toolbox is tracking and low when the Toolbox is not tracking Fig 48 describes the inter connections for the four digital I O lines between the 25 way D type connector and the digital I O lines of a ViSaGe or VSG2 5 When changing the state of a line it is good practice to allow for a transition period so that it becomes impossible for a line to be read at the precise moment that it is being set 53 Video Eyetracker Toolbox Communication Digital I O Pin on 25 way D type ViSaGe or VSG2 5 Line Connector 1 O Line 0 DOUT 6 1 DOUT 7 2 DOUT 8 3 21 DIN 7 Fig 48 Digital I O Pin Out Details Synchronisation High Speed Systems Eyetracker Interface Box Inputs The High Speed system must be used in conjunction with the Eyetracker Interface Box EIB Details of how to connect this are available in the Hardware Instal
7. 1 is designed to demonstrate basic eye tracker functionality in a Dual VGA configuration and is suitable for use with the 50 Hz and High Speed systems This example displays a single bitmap image calibrates your subject tracks subsequent eye positions for 30 seconds displays the results and finally saves the collected data to a file This demonstration does not require a CRS Visual Stimulus Generator VSG but assumes you are using a dual Windows display configuration with the Primary output set to your Windows desktop monitor and the Secondary output set to display the visual stimulus See the Dual VGA section in the Calibration Chapter for further details Run the demoVET_simpleVGAstimulus m script The following lines of code initialise the Video Eyetracker Toolbox system and the stimulus display device 58 Video Eyetracker Toolbox Communication function demoVET_simpleVGAstimulus Declare CRS MATLAB constants global CRS if isempty CRS crsLoadConstants end Set the stimulus device to Dual VGA For use with a VSG use the constant CRS deVSG vetSetStimulusDevice CRS deVGA Ask the user to select a video source errorCode vetSelectVideoSource CRS vsUserSelect if errorCode lt 0 error Video Source not selected end which will open the window shown in Fig 54 select video source select source camera Video Eyetracker Camera es ad Device Type Cancel Standard Camer
8. 106 vetSetDeviceParameters 106 82 Video Eyetracker Toolbox vetSetEIBLEDsEnabled vetSetEIB_XYOutputType vetSetFixationPeriod vetSetFixationRange vetSetMimicBackgroundColor vetSetMimicFixationColor vetSetMimicPersistence vetSetMimicPersistenceStyle vetSetMimicPersistenceType vetSetMimicScreenDimensions vetSetMimicTraceColor vetSetStimulusDevice vetSetUlMonitor vetSetViewingDistanceMM vetStartRecordingToFile vetStartTracking vetStopRecording vetStopTracking vetSetCallbackClearDisplay vetSetCallbackCollectResults vetSetCallbackDrawTarget vetSetCallbackFixate vetSetCallbacklnitialiseDisplay vetSetCallbackRegionChanged Appendices 107 108 108 109 110 110 111 111 112 113 113 114 115 115 116 116 117 117 117 118 119 119 120 120 83 Video Eyetracker Toolbox Appendices vetAddRegion Video Eyetracker Toolbox Regions Of Interest are intended to assist with fixation monitoring tasks These regions are defined as rectangular patches lying in the plane of the stimulus display There may be as many regions as you wish but no two regions may overlap Each region has a unique identifying number called the regionHandle If the subject s direction of gaze intersects a region that region s identifying number is recorded in the Video Eyetracker Toolbox data buffer along with all the other eye position data that is being collected The currently active region s RegionHandle can also be obtained using the v
9. 53 for details Data dropped A Boolean value where 0 FALSE and 1 TRUE Data trigln High Speed system only EIB Trigger input values See also vetIsEyeDataAvailable vetGetLatestEyePosition vetClearDataBuffer 92 Video Eyetracker Toolbox Appendices vetGetCalibrated This function returns True if the subject has been calibrated and False otherwise You can use the vetCalibrate or vetCalibrateAdvanced functions to calibrate your subject s direction of gaze MATLAB Syntax isCalibrated vetGetCalibrated Return Values isCalibrated A Boolean 1 x 1 matrix where 0 FALSE and 1 TRUE See also vetGetPupilCalibrated vetGetEIB_LEDSENABLED This property will determine whether the Eyetracker Interface Box LEDs are currently set on or off This property when read will return true or false accordingly MATLAB Syntax LEDsEnabled vetGetEIB_LEDsEnabled Return Values LEDsEnabled A Boolean 1 by 1 matrix where 0 FALSE and 1 TRUE See also vetVideoSourceType vetSetEIB_LEDsEnabled vetGetEIB_ XYOutputType This function will return the XY output type for the Eyetracker Interface Box The values output on the two analog channels 93 Video Eyetracker Toolbox Appendices These can either be 100 mm per volt 1000 mm per volt degrees elevation azimuth Helmholtz or degrees longitude latitude Fick This value will be one of the following constants CRS otMM100 MM 10
10. An integer scalar 1 by 1 matrix holding one of the following values CRS vsUserSelect Displays a dialog box to allow the user to select a camera type 104 Video Eyetracker Toolbox Appendices CRS vsCamera Progressive scan 50 Hz camera no identifying label on camera case CRS vsFile Takes video from a pre recorded saved CAM file CRS vsCameral Interlaced mode 50 Hz camera white Camera l label on camera case CRS vsHighSpeedCamera60 High Speed Camera running at 60Hz CRS vsHighSpeedCamera100 High Speed Camera running at 100Hz CRS vsHighSpeedCamera160 High Speed Camera running at 160Hz CRS vsHighSpeedCamera200 High Speed Camera running at 200Hz CRS vsHighSpeedCamera250 High Speed Camera running at 250 Hz If the value of this parameter is set to CRS vsUserSelect then a dialog box will appear prompting the user to select a video source This is useful if you want to distribute your software If CRS vsFile is used then a filename is expected specifying a prerecorded CAM video file CAM files can be created using the vetStartRecordingToFile and vetStopRecording functions Other constant values refer to different camera types The High Speed Video Eyetracker Toolbox camera can run at a selection of different speeds up to 250 Hz Selecting the appropriate constant here will set the camera to run at the desired speed NOTE Higher speeds can put greater demands o
11. Ensure that the camera lens and infrared mirror are free of dirt or fingerprints using the supplied lint free cloth 5 While the subject is still and viewing straight ahead adjust the height of the camera assembly so that the subject s eye appears vertically centred within the Toolbox s camera window 6 Horizontally position the subject s eye inside the camera window by adjusting the horizontal position of the camera This is achieved by releasing the locking screw located at the back of the camera assembly and sliding the camera case in the corresponding direction Make sure that the locking screw is gently tightened once the correct location is found 22 Video Eyetracker Toolbox Calibration WARNING High Speed System Only The High Speed camera can get very warm when used for extended periods The camera case acts as a heat sink for the electronics inside and therefore runs at an elevated temperature This is normal but your users should be aware of this 7 While the subject is still looking straight ahead rotate the focussing ring until the subject s pupil not their eyelashes or eyebrow appears as sharp as possible in the video image 8 Ensure that both Purkinje reflections remain distinct from any bright secondary glints This may occur if the subject is wearing glasses or if the ambient illumination contains a significant infrared component The Toolbox is designed to cope with any number of secondary glints pro
12. Eyetracker Toolbox comes complete with a demonstration program called videoEyetrace which allows you to calibrate a test subject and present simple image files If you have a suitable CRS Visual Stimulus Generator you can also use videoEyetrace to present smooth pursuit and saccade stimuli for oculomotor research These topics are covered in detail in the following 2 chapters Selected stimulus Smooth Pursuit Task vsg Fig 1 Smooth Pursuit Eye Movements 10 Video Eyetracker Toolbox Introduction Once you become familiar with the operation of your Video Eyetracker Toolbox you can exploit the Toolbox s open interface design either to e Integrate eye tracking functionality with your existing laboratory software or Write new eye tracking applications using a familiar programming environment Your Video Eyetracker Toolbox is designed to allow all aspects of its functionality to be accessed and controlled from MATLAB or any Windows programming tool that support Microsoft COM A complete description of the MATLAB interface including worked examples is covered in the final chapter of this manual A separate document describing the COM interface and examples in Delphi C Builder and Visual Basic are available to download from the Cambridge Research Systems Support Portal www crsltd com support login The 50 Hz version of the Video Eyetracker Toolbox is also compatible with Eye Response Technologies s GazeTracker
13. Fig 53 This input can be a TTL or analogue voltage signal within the range of 5 volts In the MATLAB Data results structure the values are returned in the trigIn field and are in the range of 1 This input also appears in videoEyetrace as the EIB Trig ADC graphical plot and is also logged as part of the ved results data file 55 Video Eyetracker Toolbox Communication CRS VSG 2 5 The High Speed system can receive digital timing triggers from a VSG2 5 visual stimulus generator The supplied trigger cable connects from the Digital Trig port on your EIB Fig 51 to the 25 pin Digital I O port on your Feature Connector Fig 52 or BNC Breakout Box This input can also be three TTL digital voltage signals 0 or 5 volts The trigger signals are transmitted from the EIB via its USB port to the PC where it is logged in the Digital I O column of the videoEyetrace results file and the digitallO field in the MATLAB Data structure see Appendix G for further details However unlike the 50 Hz system the High Speed version does not output a digital tracking signal to the VSG s Digital Input line 7 but is instead output on the Tracked BNC connector see Fig 53 The High Speed system can accept an input trigger from other external devices via the Trig In BNC connector on the EIB front panel see Fig 53 This input can be a TTL or analogue voltage signal within the range of 5 volts In the MATLAB Data results structure the values are retur
14. Fig 67 Tracked Video Image Calibration Geometry Calibration is required first to convert the image feature positions into a computed view direction This direction together with the computed eye position then requires further calibration to compute the corresponding gaze screen position Both halves of this calibration process are achieved by requesting the subject to view a number of screen targets from a known viewing distance Placing the ilumination sources symmetrically on either side of the camera implies that the centre of corneal curvature in each image is simply the midpoint between the resulting purkinje images 124 Video Eyetracker Toolbox Appendices This combined with requesting the subject to view a number of screen targets from a known distance which requires the user to rotate the eye through known angles enables the Toolbox to determine the 3D distance between the cornea and pupil centres This distance is denoted as K in the above illumination diagram Once K is known for a particular individual and camera setting the system can now determine the direction of gaze relative to the camera This gaze direction is denoted in the equation below as the vector Cx Cy Cz and the centre of corneal curvature as the vector Xo Yo T do boxe es BTX Y A offse T 2 d e f C x E j Y Y oet T g h lle Fig 68 Calibration Equation The remaining unknowns in the calibration equation are the
15. a PC upgrade Fig 81 250 Hz Eyetracking 144 Video Eyetracker Toolbox Appendices Noisy eye tracking measurements may be due to one of the following causes e The eye is no longer in focus due to head movement away from the infrared mirror Ensure that the subject maintains contact with the headrest at all times and refocus if necessary e Poor pupil contrast caused by room or outdoor illumination interfering with the system s infrared eye illumination This may be reduced by altering the room lighting shielding the headrest from sources of outdoor illumination or removing spectacles if necessary e Poor pupil illumination Ensure both sets of infrared LEDs are not occluded in any way and that both Purkinje glints appear compact and visible in the video image High Speed System Only The High Speed Video Eyetracker Toolbox has an adjustable iris which should always be set fully open so that the image is as bright as possible see Fig 82 Accidental adjustment of the lens iris will result in a darker image as shown in Fig 83 which will result in noisy or intermittent tracking Focussing Ring L Fig 82 Focus and Iris Rings on High Speed Camera Assembly 145 Video Eyetracker Toolbox Appendices Rum Zr Fig 83 Iris fully Open Left and partially Closed Right 146 Video Eyetracker Toolbox Notes d 147 Video Eyetracker Toolbox Notes 148 Video Eyetracker Toolb
16. also vetGetLatestEyePosition vetStartTracking vetClearMimicScreenBitmap Use this function to clear the background bitmap displayed on the mimic screen MATLAB Syntax vetClearMimicScreenBitmap See also vetCreateMimicScreen vetDestroyMimicScreen vetSetMimicScreenDimensions vetMimicPersistence vetMimicPersistenceType vetMimicTraceColor vetCreateCameraScreen This function creates a camera screen displaying live video from the eye tracking cameras providing visual feedback of pupil and Purkinje reflection tracking You should call this function just prior to calibration so that you can adjust the position and focus of the camera to obtain a high quality image of your subject s eye for tracking 88 Video Eyetracker Toolbox Appendices MATLAB Syntax vetCreateCameraScreen See also vetSelectVideoSource vetDestroyCameraScreen vetSetCameraScreenDimensions vetCreateMimicScreen This function creates the mimic screen window using the default system parameters The mimic screen is used to give visual feedback on where the subject is looking MATLAB Syntax vetCreateMimicScreen See also vetDestroyMimicScreen vetSetMimicScreenDimensions vetMimicPersistence vetMimicPersistenceType vetLoadBitmapToMimicScreen vetMimicTraceColor vetDeleteRegion This function is used to delete a specific region from the Video Eyetracker Toolbox Regions Of Interest list NOTE It is often preferable to delete all region
17. and vetSetFixationRange functions to modify the criteria that the Video Eyetracker Toolbox software uses to determine if a fixation has taken place or not MATLAB Syntax vetSetCallbackFixate callbackString 119 Video Eyetracker Toolbox Appendices Parameters callbackString A character array MATLAB string containing the callback string to execute This could for example be the name of an M file function that handles the event See also vetSetFixateRange vetSetFixationPeriod This function is used during manual calibration of a user defined stimulus device for example LED or laser galvanometer stimulus systems The specified callback function will be processed when the VET calibration routine requests that the stimulus device is initialised MATLAB Syntax vetSetCallbackInitialiseDisplay callbackString Parameters callbackString A character array MATLAB string containing the callback string to execute This could for example be the name of an M file function that handles the event See also vetSetCallbackClearDisplay vetSetCallbackDrawTarget This function allows you to specify a MATLAB string to be processed whenever the direction of gaze first enters a new Region Of Interest MATLAB Syntax vetSetCallbackRegionChanged callbackString 120 Video Eyetracker Toolbox Appendices Parameters callbackString A MATLAB string containing the callback string to execute This could
18. be fixating the fixation flag in the results data structure will be set to TRUE for the duration of the fixation The MATLAB string specified using the vetSetCallbackFixate function will be executed and the current fixation location can be returned using the vetGetFixationLocation function 109 Video Eyetracker Toolbox Appendices MATLAB Syntax vetSetFixationRange fixationRange Parameters fixationRange A positive non zero rational scalar 1 by 1 matrix This value has units of millimetres See also vetSetFixationPeriod vetSetCallbackFixate vetSetMimicBackgroundColor This function specifies the background colour of the mimic window MATLAB Syntax vetSetMimicBackgroundColor mimicBackgroundColor Parameters mimicBackgroundColor A 3 by 1 or 1 by 3 RGB colour vector each component of which is a rational scalar in the interval 0 1 1 0 0 produces 100 red 0 1 0 produces 100 green 0 0 1 produces 100 blue See also vetCreateMimicWindow vetSetMimicScreenDimensions vetSetMimicFixationColor vetSetMimicTraceColor vetSetMimicFixationColor This property sets the colour used to indicate fixations on the mimic window MATLAB Syntax vetSetMimicFixationColor mimicFixationColor 110 Video Eyetracker Toolbox Appendices Parameters mimicFixationColor A 3 by 1 or 1 by 3 RGB colour vector each component of which is a rational scalar in the interval 0 1 1 0 0 produce
19. device use vetSetDeviceParameters MATLAB Syntax vetSetStimulusDevice stimulusDevice Parameters stimulusDevice An integer value 1 by 1 matrix One of the following constants CRS deVGA CRS deVSG 114 Video Eyetracker Toolbox Appendices See also vetSelectVideoSource vetStartTracking vetSetDeviceParameters vetSetUIMonitor This function allows you to change the monitor that the Video Eyetracker Toolbox User Interface appears on For example if you are using a Windows Secondary VGA display configuration you can use this function to make the VET User Interface appear on either the primary or secondary Windows monitor MATLAB Syntax vetSetUIMonitor UIMonitor Parameters UlMonitor An integer scalar value 0 1 2 etc Valid values will depend on how many monitors you have in your system setup but most Dual Windows VGA display systems should return 0 or 1 vetSetViewingDistanceMM This function allows you to set the Viewing Distance in Millimetres that the Video Eyetracker Toolbox uses The Viewing Distance is the perpendicular distance between the centre of the stimulus display and the centre of rotation of the eye MATLAB Syntax vetSetViewingDistanceMM viewingDistanceMM Parameters viewingDistanceMM A positive integer scalar 1 by 1 matrix See also vetStartTracking vetStopTracking 115 Video Eyetracker Toolbox Appendices vetStartRecordingToFile Call vetStar
20. for example be the name of an M file function that handles the callback See also vetAddRegion vetDeleteRegion vetClearAllRegions vetRegionCount 121 Video Eyetracker Toolbox Appendices f p 122 Video Eyetracker Toolbox Appendices Illumination Geometry The Video Eyetracker Toolbox illuminates the eye using two sets of infrared diodes identically positioned on opposite sides of the infrared camera Each set of diodes will create a glint known as a Purkinje image as shown in the diagram below Fig 65 Illumination Geometry Fig 66 Typical Video Image Imaging the eye using two infrared illumination sources creates a dark pupil with two bright horizontally separated Purkinje glints as shown in Fig 66 123 Video Eyetracker Toolbox Appendices The Purkinje glints will remain bright compact and constantly separated provided they are both imaged off the corneal surface This gives allowable eye rotations of up to 40 degrees and creates even ilumination over the whole eye The Toolbox s dedicated image processing algorithms are designed to locate the pupil and both Purkinje glints in real time The algorithms exploit their size shape intensity and relative locations to optimise tracking efficiency and once located their centres are determined to sub pixel precision by ellipse fitting Tracking is indicated by superimposing the fitted ellipses onto the live eye video window as shown below
21. l as you are running an interlaced video camera If there is no label select Standard Camera as your camera is running in progressive scan mode If you have purchased the High Speed system select one of the High speed Camera options as shown in Fig 23 Alternatively if you wish to perform eye tracking on an existing video cam file select the folder icon and use the browse button to locate the relevant file Title bar The Title bar displays the program control icon the application and active protocol names and the minimise maximise and close buttons 37 Video Eyetracker Toolbox videoEyetrace Program icon Protocol name 5 Video Eyetracker toolbox New Application name Fig 24 Title Bar Menu bar The Menu bar displays the headings that lead to the menu controls These menu controls are permanently available with the exception of the Save tab which is disabled if the current data has already been saved As with most Windows programs the menu buttons are grouped together depending upon their functionality and organised under more generic menu headings file management menu buttons stimulus presentation and calibration menu buttons access options window and graph scaling menu buttons Fig 25 Menu Bar Toolbar The tool bar contains buttons that duplicate the functions of the more frequently used menu options The main purpose of the tool bar is to provide quick and easy access to these menu controls Ne
22. s eye vetCreateCameraScreen Calibrate the subject and return if the user exits before completion errorCode vetCalibrate if errorCode 0 error Calibration not completed end vetSetDeviceParameters CRS deUser Check we have dual VGA monitor setup Calculate primary and secondary monitor positons MonitorPos get 0 MonitorPosition if numel MonitorPos lt 8 error Less than two monitors detected end PrimaryPos MonitorPos 1 SecondaryPos MonitorPos 2 PrimarySize PrimaryPos 3 PrimaryPos 1 PrimaryPos 4 PrimaryPos 2 1 SecondarySize SecondaryPos 3 SecondaryPos 1 SecondaryPos 4 SecondaryPos 2 1 PrimaryLowerLeft PrimaryPos 1 T SecondarySize 2 PrimaryPos 2 PrimarySize 2 2 SecondaryLowerLeft SecondaryPos 1 PrimarySize 2 SecondaryPos 2 PrimarySize 2 2 Display a dialog to allow the user to select the stimulus display that they wish to use MonitorToUse uigetpref Monitors One Stimulus Monitor Selection Dialog Select which monitor to use 1 2 sss CheckboxString Always use this monitor switch MonitorToUse case 1 SelectedPos PrimaryLowerLeft PrimarySize case 2 SelectedPos SecondaryLowerLeft SecondarySize end 66 Video Eyetracker Toolbox Communication Pick an image file to display and get the full path to the file The ima
23. used in the calibration The total number of targets used for calibration is the product of these to settings Screen Scale Factor This parameter determines the percentage of the screen originating from the centre that the targets will be displayed over Setting this value to 100 would result in the targets covering 100 of the visible screen area A setting of 50 would result in the targets covering 50 of the total screen area Shape Allows different shapes for the target to be selected This setting also enables a bitmap image such as a cartoon character or face to be used as the calibration target Size Sets the size of the calibration targets This setting does not apply if a bitmap image has been selected as the target Target Colour Specifies the colour of the calibration targets Again this setting does not apply if a bitmap image is being used Background Colour Set the background colour that the targets are presented on Filename If a bitmap image has been selected in the Shape setting this parameter will become active and will allow you to browse specify the bitmap file to be used as the calibration target Note All of these parameters except the parameters located in the Calibration Points group box will be ignored if you have elected to use a non computer based device as the presentation display The appearance of this display and calibration targets are left entirely to the user when working with Custom Devic
24. 0 mm V CRS otMM1000 MM 1000 mm V CRS otHelm1 0 1 Volts per degree Helmholtz coordinates CRS otFicks1 0 1 Volts per degree Fick coordinates MATLAB Syntax XYOutputType vetGetEIB XYOutputType Return Values XYOutputType An integer constant which will be one of the values listed above See also vetGetLatestEyePosition vetVideoSourceType vetSetEIB XYOutputType vetGetFixationLocation This function returns the location of the current fixation This function will return 0 0 0 if the subject is not fixating MATLAB Syntax isFixating xPositionMM yPositionMM vetGetFixationLocation Return Values isFixating A logical value 1 by 1 matrix containing the value 0 if the subject is not fixating or 1 if the subject is fixating xPositionMM The x position in mm of the fixation or 0 if the subject is not fixating yPositionMM The y position in mm of the fixation or 0 if the subject is not fixating 94 Video Eyetracker Toolbox Appendices See also vetGetLatestEyePosition vetGetLatestEyePosition This function returns the most recent eye position analysed by the Video EyeTracker Toolbox This data point is NOT removed from the data buffer that the Video Eyetracker Toolbox maintains l e you can get the same results later using vetGetBufferedEyePositions If there are no points available in the buffer a MATLAB structure filled with zeros is returned together with a non zero 1 Err
25. A v7 bd F A Fig 29 Zoom Icons Status bar The Status bar displays information about videoEyetrace The first section of the status bar displays the name of the currently selected stimulus and the second section displays the name of the data file currently being shown on in the data window The second section of the status bar will stay empty until either data has been collected or a data file has been opened currently selected stimdus Selected Stimulus Saccade Task nare d curert cata file Fig 30 Status Bar Changing Stimuli and Settings The presentation stimulus can be controlled or modified using the Stimulus window This can be accessed by either clicking on the Mona Lisa icon on the Toolbar or by clicking View Stimulus from the Menu bar Fig 31 videoEyetrace Stimulus Icon 40 Video Eyetracker Toolbox videoEyetrace Video Eyetracker Options stimulus Global Parameters Background Coo ST C Stimulus Selection Stimulus Stil Picture vga Fig 32 videoEyetrace Stimulus Window The window shown in Fig 32 controls the type of stimulus presented to the subject The currently selected stimulus is displayed in the stimulus combo box located in the Stimulus Selection area of the tab Different stimuli can be chosen by clicking on the down arrow to the right of the combo box and selecting a new stimulus from the list Fig 32 shows that there are 3 stimulus options available if you are using a VSG
26. B matrix containing the mimic window bitmap See also vetCreateMimicScreen vetLoadBitmapToMimicScreen vetGetPupilCalibrated This function returns True if the pupil size has been calibrated False otherwise The vetCalibrate or vetCalibrateAdvanced functions can be used to calibrate your subject s pupil size MATLAB Syntax isPupilCalibrated vetGetPupilCalibrated Return Values isPupilCalibrated A Boolean 1 x 1 matrix where 0 FALSE and 1 TRUE 96 Video Eyetracker Toolbox Appendices See also vetGetCalibrated vetGetRegionCount The vetGetRegionCount function returns the number of Regions Of Interest that are currently defined in the Video Eyetracker Toolbox Regions Of Interest list MATLAB Syntax RegionCount vetGetRegionCount Return Values regionCount A non negative zero included integer scalar 1 by 1 matrix See also vetAddRegion vetDeleteRegion vetClearAllRegions vetGetResultsCount The Video Eyetracker Toolbox maintains a buffer of eye position data The vetGetResultsCount function returns the number of results currently stored in this buffer MATLAB Syntax resultsCount vetGetResultsCount Return Values resultsCount A non negative zero included integer scalar 1 by 1 matrix See also vetClearDataBuffer vetGetLatestEyePosition vetlsEyeDataAvailable 97 Video Eyetracker Toolbox Appendices vetGetTracking This function will return True or False de
27. DataBuffer This will create a comma delimited results data file which can be imported into other applications Fig 59 shows the data when it has been imported into Microsoft Excel You can see all the calibration parameters plus the full eye tracking results for every processed image 64 Video Eyetracker Toolbox Communication Tem Re 154 Ble Edt View Insert Format Tools Data RoboPDF Window Help forhep BX inu i 1228 z 21 e give Heil 3 R61 E A 8 E D E F G H peg 5 N_ 1 EyeTracker Toolbox version 3 02 2 CalibrationParameters A11 0 05229 3 CalibrationParameters A12 0 36173 4 CalibrationParameters A13 0 01235 5 CalibrationParameters A21 0 44936 5 CalibrationParameters A22 0 03279 7 CalibrationParameters A23 0 12607 8 CalibratinParameters A31 0 00016 9 CalibrationParameters A32 0 00002 10 CalibrationParameters E11 0 18003 11 CalibrationParameters E12 0 00005 12 CalibrationParameters E21 0 0001 13 CalibrationParameters E22 0 20009 14 CalibrationParameters X 15 0971 15 CalibrationPararneters_Y 2 00868 16 CalibrationParameters Rpc 147 21 17 CalibrationParameters Mean 185 837 18 CalibralionParameters MeanY 137 334 19 CalibrationParameters ViewDist 1000 20 TimeStamp Tracked Calibrated ScreenPoc ScreenPos FicksLong FicksLat Helmholtz Helmholtz PupilDi
28. Examples vetLoadCalibrationFile C Temp SubjectA scf will open the file SubjectA scf located in the temp folder stored on the C drive vetLoadCalibrationFile SubjectA scf will fail to open because the full path to the file has not been specified See also vetCalibrate vetCalibrateAdvanced vetSaveCalibrationFile 101 Video Eyetracker Toolbox Appendices vetLoadBitmapToMimicScreen Use this function to load a background bitmap onto the mimic screen The mimic screen is a graphical representation of eye location If a scaled image of the stimulus or of landmarks in the stimulus is specified by the experimenter then the mimic screen will display the subject s gaze with respect to landmarks in the stimulus image MATLAB Syntax vet LoadBitMapToMimicScreen ImageMat rix ScaleToStimulusDevice Parameters ImageMatrix Specifies the image values to be loaded onto the mimic window This should be either a 3 dimensional width by height by 3 matrix 1 page each for Red Green and Blue or a 2 dimensional width by height matrix for a monochrome image The sample values in this matrix should lie in the range 0 1 ScaleToStimusDevice If this is TRUE 1 the bitmap is scaled to the same scale as the selected stimulus device otherwise if it is FALSE 0 it is stretched to fill the whole of the mimic screen Note Using the vetSetDeviceParameters function to set the presentation device to CRS deUser resu
29. Hardware states after Subject Calibration CRS Visual Stimulus Generator In order for the VSG to perform a subject calibration it is necessary for the Video Eyetracker Toolbox to draw targets onto some of the VSG s video pages The Toolbox also needs to modify the VSG s palette and may change global settings such as the spatial units and colour space that are being used The following section describes the state of a VSG after a subject calibration has been performed using the Video Eyetracker Toolbox with the CRS Toolbox for MATLAB For ease of use we recommend that you try to draw and define your presentation stimuli after you have run a subject calibration or if this is not possible then you should be aware that some of the video pages that your presentation stimulus appears on may be corrupted or cleared and that the spatial units colour spaces and Object Animation System objects you have previously defined may also need to be reset e The first and second video pages will be cleared using pixel level 254 CRS BACKGROUND The Object Animation System OAS will be turned off All OAS objects created prior to running a subject calibration will be destroyed Overlay pages will not be visible Spatial units will be set to CRS PIXELUNIT e The top left corner of video memory space will be set to 0 0 e The drawing origin will be set to the centre of the screen The active colour space will be set to CRS CS_RGB
30. Refer back to the subject calibration section for further details Fig 78 Excessive View Angle e The image contains a shadow or foreign object that creates a region darker than the pupil itself 141 Video Eyetracker Toolbox Appendices System fails to track the Purkinje Glints Failure of the system to locate both Purkinje glints could be due to one or more of the following e The pupil is no longer fully contained within the image The image contains a spurious bright glint which is lies close to the pupil This may be caused by certain types of spectacles or from secondary sources of infrared illumination Fig 79 Spurious Additional Glints e The eye is no longer in focus due to head movement away from the infrared mirror see Fig 76 e One of the Purkinje glints has merged with a spurious glint and is being rejected on position size shape criteria This is often the result of the eye rotating to the point where one of the Purkinje glints no longer lies on the corneal surface see Fig 78 142 Video Eyetracker Toolbox Appendices Eye tracking appears inaccurate Occasionally there may be consistent errors between where the subject is looking and where the system thinks the subject is looking This may be solved by checking the following The eye may be slightly out of focus due to head movement away from the infrared mirror Ensure that the subject maintains contact with the headrest at a
31. UE otherwise it is 0 FALSE The diameter of the test subject s pupil in millimetres Direction of gaze projected onto stimulus display in mm X Y Eye rotation in Fick coordinates Longitude Latitude Eye rotation in Helmholtz coordinates Elevation Azimuth Current region number A value containing the combined digital values see page 53 for details High Speed system only A 5 volts TTL or analogue voltage input signal which is returned in the range of 1 136 Video Eyetracker Toolbox Appendices d Appendix H Troubleshooting In this section System Slow to Calibrate 139 System Fails to Calibrate 139 System fails to track the Eye 139 System fails to track the Pupil 140 System fails to track the Purkinje Glints 142 Eye tracking appears inaccurate 143 Eye tracking appears noisy erratic or intermittent 143 137 Video Eyetracker Toolbox Appendices Poor performance or complete eye tracking failure is likely to be the result of poor operating conditions rather than a fault with the Toolbox itself The troubleshooting guide is designed to eliminate those problems which are most often experienced by new users If after making full use of this guide you are still experiencing problems please create a Ticket using the Cambridge Research Systems support postal www crsltd com support login The following sections use example video images from the 50 Hz system although the issues ap
32. Video Eyetracker Toolbox d Video Eyetracker Toolbox User Manual version 3 11 January 2006 Cambridge Research Systems Ltd 80 Riverside Estate Sir Thomas Longley Road Rochester Kent ME2 4BH England WWW crsltd com Video Eyetracker Toolbox Version Date Changed by Description of Changes 3 11 18 Jan 06 M Hodgetts New sections for High Speed S Elliott Video Eyetracker Toolbox Revised videoEyetrace description Added MATLAB worked examples Appropriate commands from the CRS Toolbox for MATLAB described in the Appendix Description of COM Server interface removed Video Eyetracker Toolbox Contents d Introduction 9 Overview of Video Eyetracker Toolbox Family 10 CRS Video Eyetracker Toolbox 12 50 Hz Video Eyetracker Toolbox 12 High Speed Video Eyetracker Toolbox 12 Images Acquired with 50 Hz and High Speed Systems13 Calibration 15 Monitor Configuration and Calibration 16 Dual VGA 16 Dual VGA with GazeTracker 18 CRS Visual Stimulus Generator 19 Non Computer based Custom Display Device 20 Positioning and Focusing your Subject Imaging 21 Subject Calibration 24 Pupil Scale Calibration 25 Subject Gaze Calibration 27 Hardware states after Subject Calibration 33 CRS Visual Stimulus Generator 33 Dual VGA 34 Custom Device 34 Video Eyetracker Toolbox Contents videoEyetrace 35 Overview 36 The videoEyetrace Window 36 Initialising the Video Eyetracker Toolbox Camera 37 T
33. a Fig 54 Camera Movie File Source Window Select the Video Eyetracker Camera which corresponds to your supplied system from the Device Type drop down box This option should be highlighted as shown in Fig 54 but if not click the EyeLock button to its left You can select from Standard Camera Standard Camera l or one of the High Speed options Once the correct camera has been selected click OK to run the following lines of code which open up the Camera and Calibration windows oe Create a live camera screen so that we can position the camera to image the test subject s eye oe vetCreateCameraScreen 59 Video Eyetracker Toolbox Communication Calibrate the subject and return if the user exits before completion ole errorCode vetCalibrate if errorCode lt 0 error Calibration not completed end Clear the VET VGA calibration window vetSetDeviceParameters CRS deUser Once your subject is correctly positioned click Start in the Calibration window The calibration routine will now proceed by requesting the test subject to fixate on a number of calibration targets on the stimulus monitor Subject Calibration VET Camera Window B Gy calibrating step 5 of 20 Sy VET Camera Window 9 Advanced Pupil Target Diameter 2 Save as Default Calibration Fig 55 Live Camera and Calibration Windows Calibration in Progress Once calibration is compl
34. am Region Fixation DigitallO Dropped 21 0 1 981 9743 056 056 056 056 210 1 o o 0 2 20 1 1 10975 312 063 053 063 053 208 1 o 0 o 23 40 1 1 10791 943 062 054 062 054 209 4 o o 0 24 60 1 1 11087 9334 063 053 063 053 209 4 D 0 25 80 1 1 100011 9 446 057 754 057 054 212 A D o 0 26 400 1 1 10997 9 369 063 054 053 054 21 4 D 0 0 27 120 1 1 8474 562 049 032 049 032 200 4 o 0 0 28 140 1 1 8385 5454 021 031 021 031 215 A o 0 0 28 160 1 1 4102 523 024 03 024 03 27 A D o 0 30 180 1 1 5642 5891 032 034 032 034 211 4 D o 0 31 200 1 1 5608 5 941 032 034 032 034 21 A o 0 0 1 1 5413 aga nal nal 2n 1 n na M 4 9 H N myResults Ie x Ready NUM Fig 59 MATLAB Eyetracking Results Data File The full code listing for this MATLAB example is as follows function demoVET simpleVGAstimulus This demonstrates the most basic functionality of the Video Eyetracker Toolbox oe oo Declare CRS MATLAB constants global CRS if isempty CRS crsLoadConstants end Set the stimulus device to Dual VGA For use with a VSG use the constant CRS deVSG vetSetStimulusDevice CRS deVGA Ask the user to select a video source errorCode vetSelectVideoSource CRS vsUserSelect 65 Video Eyetracker Toolbox Communication if errorCode lt 0 error Video Source not selected end Create a live camera screen so that we can position the camera to image the test subject
35. and a Still Picture option if you are in dual VGA mode Once the required stimulus has been selected its parameters can be edited by clicking on the Properties button to the right of the Stimulus combo box The background colour used for the stimuli is displayed in the colour panel remember this colour is not gamma corrected and can be edited by either double clicking in the colour panel or clicking on the browse button to the right of the panel Changing View Options Click on the View Options icon on the toolbar Fig 33 videoEyetrace View Options Icon 41 Video Eyetracker Toolbox videoEyetrace to display the window shown in Fig 34 Video Eyetracker Options options Graph Configuration Mimic Window EIB Results Results Type Screen Position mm Appearance Horizontal position mm em Vertical position rm Untracked warning Gass EIB Trig ADC Pupil diameter Digital 1 0 t E EJ EJ Fig 34 Video Eyetracker Options Graph Tab Once the stimulus has been selected the Options window contains everything else necessary to configure your experimental setup This window has four different tabs Graph Configuration Mimic Window and EIB Graph Tab Clicking on the Graph tab see Fig 34 will reveal the following options The Results sub window allows videoEyetrace to return the viewed screen position either in millimetres or in Fick or Helmholtz angle co ordinates see Append
36. andles the event See also vetlsEyeDataAvailable vetGetBufferedEyePosition vetGetLatestEyePosition 118 Video Eyetracker Toolbox Appendices This function is used during manual calibration of a user defined stimulus device for example LED or laser galvanometer stimulus systems The specified callback function will be processed when the VET calibration routine requests that a target be drawn MATLAB Syntax vetSetCallbackDrawYarget callbackString Parameters callbackString A character array MATLAB string containing the callback string to execute This could for example be the name of an M file function that handles the event This function will be called with two parameters the X and Y locations of the target to be drawn in mm so for example if callbackString is specified as myCallbackFunction it will be called as myCallbackFunction xPosMM yPosMM See also vetCalibrate vetCalibrateAdvanced vetSetDeviceParameters vetSetCallbackClearDisplay vetSetCallbacklnitaliseDisplay This function allows you to specify a MATLAB string to be processed whenever the subject makes a fixation Typically this string will contain the name of a MATLAB M file function This is called the callback function Callback functions are typically designed to execute rapidly If a second fixation occurs before the first callback has finished it s processing that fixation will be ignored You can use the vetSetFixationPeriod
37. articular location before the Toolbox considers it to be a fixation When a subject is found to be fixating the fixation flag in the results data sets will be set to 108 Video Eyetracker Toolbox Appendices TRUE The value specified will be rounded UP to the nearest multiple of the frame time 20ms for 50 Hz 4ms for 250 Hz The minimum fixation period that can be specified is 100 ms and the maximum fixation period that can be specified is 60000 ms When a subject is found to be fixating the fixation flag in the results data structure will be set to TRUE for the duration of the fixation The MATLAB string specified using the vetSetCallbackFixate function will be executed and the current fixation location can be returned using the vetGetFixationLocation function MATLAB Syntax vetSetFixationPeriod fixationPeriod 7 Parameters fixationPeriod A positive non zero integer scalar 1 by 1 matrix in the interval 100 60000 This value has units of milliseconds See also vetSetFixationRange vetSetCallbackFixate vetSetFixationRange Use the vetSetFixationRange function to specify the fixation tolerance range in millimetres This value is used in the calculations that determine if a subject is fixating A subject is deemed to be fixating if his gaze does not move by more than the Fixation Range in any one Fixation Period The Fixation Period is set using the vetSetFixationPeriod function When a subject is found to
38. been successful otherwise will return False 0 If you wish to calibrate a subject without passing any parameters you should use the Video Eyetracker Toolbox vetCalibrate function instead MATLAB Syntax calibrationSuccessful vetCalibrateAdvanced xPoints yPoints targetShape targetSize targetColour backgroundColour ScaleFactor fixationDuration accuracyLevel imageFilename Parameters numxX A scalar integer value in the closed interval 2 10 indicating the number of dots across to use numY A scalar integer value in the closed interval 2 10 indicating the number of dots down to use targetShape A scalar integer value drawn from the following set CRS tsSquare CRS tsCircle CRS tsCross CRS tslmage targetSize A rational scalar value in the closed interval 0 01 5 0 This indicates the diameter of the target measured in degrees of Arc This value is ignored when a tslmage is selected targetColour A 3 by 1 or 1 by 3 RGB colour vector each component value a rational number in the interval O 1 backgroundColour A 3 by 1 or 1 by 3 RGB colour vector each component value a rational number in the interval 0 1 scaleFactor A scalar value in the interval 1 100 indicating the calibration area as a percentage of total screen area a sensible value for this parameter would be 90 86 Video Eyetracker Toolbox Appendices fixationDuration A scalar value measured in milliseconds indicat
39. box maintains a data buffer in which it records eye position data The vetSaveResults function saves the contents of the data buffer to a user specified file Several different file formats are supported although it is recommended that users use either the default CRS ffCommaDelimitedNumeric format or CRS ffMATfile format The function utilVETreadDataFile is provided to read Comma Separated Value CSV files back into the MATLAB workspace 103 Video Eyetracker Toolbox Appendices MATLAB Syntax vetSaveResults filename vetSaveResults filename format Parameters Filename This specifies the name of the file which the results data will be saved to This string must include the full path eg C myDirectory results csv Format Use this flag to specify the format that the file will be written in Accepted values include CRS ffCommaDelimitedNumeric CRS ffSemiColonDelimitedNumeric CRS ffMATfile is omitted CRS ffCommaDelimitedNumeric is used by efault See also vetClearDataBuffer vetGetLatestEyePosition vetlsEyeDataAvailable vetSelectVideoSource vetSelectVideoSource can be used to select the video source used by the Video Eyetracker Toolbox vetSelectVideoSource can display a dialog box were the user can select the video source at run time or alternatively it can directly specify the camera type or pre recorded CAM file name MATLAB Syntax vetSelectVideoSource source filename Parameters Source
40. bust operation over a wide range of subjects and will easily meet its quoted technical specification if used with care It is therefore important that your subject s eye is imaged under the best possible conditions This will be achieved by carrying out the following instructions 1 Ensure that the surface of the stimulus display is both centred and perpendicular to this subject s view axis 2 Position the subject s head so that their chin is resting on the chin rest and their forehead is resting against the head strap The height of the chin rest and head strap should be adjusted so that the subject is comfortable and their eye is level with the centre of the presentation display as shown below Note the subject s head should not be tilted as this may cause the eye lid to obscure the pupil Fig 10 Correct Viewing Geometry 21 Video Eyetracker Toolbox Calibration WARNING When adjusting the height of the chin rest over large distances it is advisable to use one hand to rotate the adjuster knob and the other hand to guide the other end of the chin rest in the desired direction see Fig 11 This also applies to moving the camera assembly Support and guide this end of the chin rest when moving the vertical position over a 3 large range Fig 11 Adjusting the Chin Rest 3 Use the MATLAB calibration scripts see page 24 or another application that uses the Toolbox s functionality 4 Remove the camera lens cover
41. cedure will need to be repeated if the screen resolution of the second monitor is subsequently changed in any way 17 Video Eyetracker Toolbox Calibration Now click the Advanced button to reveal the dialogue box shown in Fig 7 VGA Calibration advanced settings Calibration Square Size 1100 w pixels Default Viewing Distance po mm Switch Displays RES Cancel OK Fig 7 VGA Viewing Distance Calibration To complete VGA Calibration it is necessary to measure and enter the correct Viewing Distance between the test subject and the stimulus monitor This window also gives you the option of changing the size of the white calibration square to 200 or 400 pixels This is sometimes preferable if working with high resolution monitors or if you are concerned with parallax measurement errors If this is changed to a 200 or 400 pixel white square it is important to measure and enter its new size as shown in Fig 6 Dual VGA with GazeTracker GazeTracker operates in a different mode to the conventional Dual VGA setup in that the Video Eyetracker Toolbox User Interface controls e g Calibration and Camera windows and the GazeTracker program run on the Secondary monitor while the VET calibration targets and stimuli presented by GazeTracker run on the Primary monitor To achieve this click the Switch Displays button on the VGA Pixel Calibration Advanced Settings as shown in Fig 7 and then complete the VGA calibration
42. cking Tidy up by clearing away the camera screen vetDestroyCameraScreen vetDestroyMimicScreen 71 Video Eyetracker Toolbox Communication VET Mimic Window VET Mimic Window Fig 62 Mimic Window during Saccade Test Finally the tracked eye positions and triggering results are plotted and the data is saved to MATLAB file Retrieve the recorded eye positions remove false DATA vetGetBufferedEyePositions remove Display the eye positions and triggers g figure plot DATA ficksPositions 1 b longitude in blue hold on plot DATA ficksPositions 2 r latitude in red hold on plot DATA digitallO g triggers in green grid on Finally save the results to disk as a mat file tempfile C myResults mat vetSaveResults tempfile CRS ffMATfile 72 Video Eyetracker Toolbox Communication Figure 1 File Edit Yiew Insert Tools Desktop Window Help eH kaana E DE Fig 63 Tracked Eye Data for Triggered Saccade Stimulus The blue and red traces in Fig 63 represent horizontal and vertical angular eye movements while the green trace represents the triggers output by the ViSaGe to coincide with the presentations of saccade targets Furthermore opening the myResults mat file in MATLAB s current directory will declare all your results as variables in MATLAB s own workspace as shown in Fig 64 73 Video Eyetracker Too
43. coefficients a to h alpha to delta and Xoffset Yoffset These are given initial estimates which are then refined by requesting the subject to view a set of screen targets with known positions Fig 69 Calibration Targets 125 Video Eyetracker Toolbox Appendices The Toolbox computes the differences between where these targets are and where the Toolbox calculates them to be using the initial estimates These differences are used to create an error term which is then minimised by altering the calibration parameters using iterative numerical techniques Those values which minimise this error term are then used in the above calibration equation for all subsequent analysis The Toolbox can now determine the position and orientation of the eye relative to the monitor and has all the information necessary to determine the view axes for all subsequent images It is therefore able to determine the current point of fixation on the monitor from the relative positions of the pupil and Purkinje centres in subsequent images 126 Video Eyetracker Toolbox Appendices p Appendix Template for Fixed Mount Eyelock Headrest 127 Video Eyetracker Toolbox Appendices The fixed mount EyeLock headrest is designed to be bolted directly to a CRS Experimental Workbench or table top for extra rigidity This requires drilling two 20 mm diameter holes into the top of the workbench to a depth of at least 20mm followed by two concentric 8 1
44. d MinTime rand MaxTime MinTime Specify target locations from 5 to 10 degrees We invert the x and y locations because we are moving the video window offset not the video memory Xlocations crsDegreesToPixels 5 0 7 5 10 0 Ylocations crsDegreesToPixels 0 0 0 0 0 0 70 Video Eyetracker Toolbox Communication Generate random locations for each entry Indices ceil rand size Pages numel Xlocations Xlocations Xlocations Indices Ylocations Ylocations Indices Make sure the fixation point always appears at 0 0 Xlocations Pages FixationPage 0 Ylocations Pages FixationPage 0 Fig 61 Fixation and Saccade Targets The next section of code defines triggers and associates them with the relevant video pages We then commence eye tracking and present the stimuli The Camera and Mimic screens are destroyed when the test is complete Specify triggers to identify when a target shift has occurred The VET captures digital output lines 6 7 and 8 Triggers zeros size Pages Triggers Pages FixationPage bin2dec 011100000 oe Set up RTS page cycling Repeating false crsSimplePageCycle Pages Xlocations Ylocations Times T riggers Repeating 7 Start eye tracking and stimulus cycling vetStartTracking crsSetCommand CRS CYCLEPAGEENABLE pause crsFramesToSeconds sum Times crsSetDisplayPage BlankPage vetStopTra
45. e chessboard will be replaced with a white square on the stimulus display To calibrate this display you must measure the width and height of the square using a ruler and if necessary adjust the horizontal or vertical scales on the display to ensure that both measurements are equal Please note that the The Size of Square box gives you the option of changing from a 100 pixel to either a 200 or 400 pixel white square which will be easier to measure and will reducing parallax errors When the white stimulus is truly square enter its height width in the Height of Square box To complete calibration it is also necessary to measure and enter the correct Viewing Distance between the test subject and the stimulus monitor in the Default Viewing Distance box Note although you can set the viewing distance parameter in your VSG program the Toolbox s calibration routines will use the value you 19 Video Eyetracker Toolbox Calibration set here to create the calibration targets and determine your subject s direction of gaze Full details of how to configure the stimulus monitor are available by clicking on the vsgDesktop Documentation plug in as shown below Note that the VSG calibration procedure will need to be repeated if the screen resolution of the stimulus monitor is subsequently changed in any way TEN MENEE z 6 p Auto Configuration Description Functions Function Dest
46. e horizontal azimuth component These coordinate systems are demonstrated in the figure below Fick Coordinates Helmholtz Coordinates Fig 73 Angular Coordinate Systems It is important to be aware of which angular coordinate system is being used as their angular components do not commute For example a longitude of 50 degrees followed by a latitude of 30 degrees is not equivalent to an elevation of 30 degrees followed by an azimuth of 50 degrees 134 Video Eyetracker Toolbox Appendices Appendix G Data Returned hy the Toolbox 135 Video Eyetracker Toolbox Technical Specification Appendices The table below contains a list of all of the entries stored in the MATLAB Data structure which is returned by the functions vetGetLatestEyePosition and vetGetBufferedEyePositions Data timeStamps Data tracked Data calibrated Data fixation Data dropped Data pupilDiameter Data mmPositions Data ficksPositions Data helmholtzPositions Data region Data digitallO Data trigIn The time in milliseconds since tracking started If the frame is successfully tracked this value is set to 1 TRUE otherwise it is 0 FALSE If the subject has been calibrated this value is set to 1 TRUE otherwise it is 0 FALSE If the subject is fixating this value is set to 1 TRUE otherwise it is 0 FALSE If the frame has been dropped not processed by the Toolbox in order to keep real time it is set to 1 TR
47. e stimulus or of landmarks in the stimulus is specified by the experimenter then the mimic screen will display the subject s gaze with respect to landmarks in the stimulus image MATLAB Syntax vet LoadBmpFileToMimicScreen ImageMatrix ScaleToStimulusDevice Parameters ImageMatrix Specifies the image values to be loaded onto the mimic window This should be either a 3 dimensional width by height by 3 100 Video Eyetracker Toolbox Appendices matrix 1 page each for Red Green and Blue or a 2 dimensional width by height matrix for a monochrome image The sample values in this matrix should lie in the range 0 1 scaleToStimulusDevice If this is TRUE 1 the bitmap is scaled to the same scale as the selected stimulus device otherwise if it is FALSE 0 it is stretched to fill the whole of the mimic screen See also vetCreateMimicScreen vetLoadBitmapToMimicScreen vetLoadCalibrationFile This function loads a Subject Calibration File previously created by the Toolbox Subject Calibration files are identified by a scf file extension and must be referenced by their absolute address vetLoadCalibrationFile will return TRUE 1 if it is successful FALSE 0 if it is unable to open the file MATLAB Syntax vetLoadCalibrationFile filename Parameters Filename A char array matlab string specifying the name of the calibration file to be opened You need to include the full windows path to the file
48. emain in the VET data buffer and will be available for reading using the vetGetBufferedEyePositions function MATLAB Syntax vetStopTracking See also vetStartTracking This function is used during manual calibration of a user defined stimulus device For example LED or laser galvanometer stimulus systems The specified callback function will be executed when the VET calibration routine requests the stimulus display to be cleared MATLAB Syntax vetSetCallbackClearDisplay callbackString 117 Video Eyetracker Toolbox Appendices Parameters callbackString A character array MATLAB string containing the callback string to execute This could for example be the name of an M file function that handles the event See also vetCalibrate vetCalibrateAdvanced vetSetDeviceParameters vetSetCallbackDrawTarget vetSetCallbacklnitaliseDisplay This function allows you to specify a MATLAB string to be processed whenever a new batch of results is ready to be collected This MATLAB string normally defines the name of a MATLAB M file which will execute when the callback occurs This script should be designed to execute quickly as any subsequent calls that collide may be dropped and lost MATLAB Syntax vetSetCallbackCollectResults callbackString Parameters callbackString A character array MATLAB string containing the callback string to execute This could for example be the name of an M file function that h
49. es Information regarding where targets should be displayed are returned in the Toolbox s vetSetCallbacklnitialiseDisplay 31 Video Eyetracker Toolbox Calibration vetSetCallbackClearDisplay and vetSetCallbackDrawTarget event call back procedures The final aspects of the Calibration window are the Test Load and Save buttons The Test button can be clicked after the calibration procedure has been completed and via visual feedback can be used to test the accuracy of a calibration In Test mode the Toolbox will display a fixed set of 9 dots and will display the subject s calculated gaze position as a red dot in the Calibration window in real time Subject Calibration Sy tracking test C Show Tracking Location on Stimulus Display Save as Default Calibration Fig 21 Calibration Window in Test Mode The Save and Load buttons allow the calibration settings for a specific subject to be stored and retrieved at any time The data is stored as a scf Subject Calibration File which should remain accurate provided it is used by the same subject that created it and the viewing geometry has not been altered in any way If in any doubt the calibration procedure should be repeated The Toolbox will continue to use the current calibration settings until the end of the session or it is explicitly overwritten either by loading a different calibration file or via recalibration 32 Video Eyetracker Toolbox Calibration
50. etGetActiveRegion function When the direction of gaze first enters a region the MATLAB function specified using vetSetCallbackRegionChanged is executed MATLAB Syntax regionHandle vetAddRegion leftMM topMM rightMM bottomM Parameters LeftMM A rational scalar 1 by 1 matrix specifying the left edge of the region of interest TopMM A rational scalar 1 by 1 matrix specifying the top edge of the region of interest RightMM A rational scalar 1 by 1 matrix specifying the right edge of the region of interest BottomMM A rational scalar 1 by 1 matrix specifying the bottom edge of the region of interest Return Values regionHandle An integer scalar 1 by 1 matrix denoting the region created by the operation Example regionHandle vetAddRegion 50 50 100 100 will add a square Region of Interest 50mm in size with a mid point 75mm right and 75mm up from the centre of the screen 84 Video Eyetracker Toolbox Appendices See also vetDeleteRegion vetClearAllRegions vetRegionCount vetSetCallbackRegionChanged vetCalibrate This function launches a calibration dialog that can be used to configure and control subject calibration Through this dialog eye position and pupil diameter can be calibrated and calibration options can be configured Normally calibration options are recalled from the previous session If specific calibration options are required they can be specified
51. etGetVideoSourceType 99 vetlsEyeDataAvailable 100 vetLoadBmpFileToMimicScreen 100 vetLoadCalibrationFile 101 vetLoadBitmapToMimicScreen 102 vetSaveCalibrationFile 102 vetSaveMimicScreenBitmap 103 vetSaveResults 103 vetSelectVideoSource 104 vetSetCameraScreenDimensions 106 vetSetDeviceParameters 106 vetSetEIB LEDsENABLED 107 vetSetEIB XYOutputType 108 vetSetFixationPeriod 108 vetSetFixationRange 109 vetSetMimicBackgroundColor 110 vetSetMimicFixationColor 110 vetSetMimicPersistence 111 vetSetMimicPersistenceStyle 111 vetSetMimicPersistenceType 112 vetSetMimicScreenDimensions 113 vetSetMimicTraceColor 113 Video Eyetracker Toolbox Contents vetSetStimulusDevice 114 vetSetUlMonitor 115 vetSetViewingDistanceMM 115 vetStartRecordingToFile 116 vetStartTracking 116 vetStopRecording 117 vetStopTracking 117 vetSetCallbackClearDisplay 117 vetSetCallbackCollectResults 118 vetSetCallbackDrawTarget 119 vetSetCallbackFixate 119 vetSetCallbacklnitialiseDisplay 120 vetSetCallbackRegionChanged 120 Appendix B Imaging Geometry 122 Illumination Geometry 122 Illumination Geometry 123 Calibration Geometry 124 Appendix C Template for Fixed Mount EyeLock Headrest 127 Appendix D Technical Specification 129 Technical Specifications 130 Appendix E Hot Mirror Spectral Transmission 131 Appendix F Fick and Helmholtz Coordinates 133 Technical Description 134 Appendix G Data Returned by the Toolbox 135 Technical Specification 135 Techn
52. ete click OK in the subject Calibration window The following lines of code will check that Windows has been configured in dual VGA mode and calculate both monitor positions Check we have dual VGA monitor setup Calculate primary and secondary monitor positions MonitorPos get 0 MonitorPosition if numel MonitorPos 8 error Less than two monitors detected endj 60 Video Eyetracker Toolbox Communication PrimaryPos MonitorPos 1 SecondaryPos MonitorPos 2 PrimarySize PrimaryPos 3 PrimaryPos 1 PrimaryPos 4 PrimaryPos 2 1 SecondarySize SecondaryPos 3 Secondary Pos 1 SecondaryPos 4j Secondatybos 27i 1 PrimaryLowerLeft PrimaryPos 1 SecondarySize 2 PrimaryPos 2 PrimarySize 2 2 SecondaryLowerLeft SecondaryPos 1 PrimarySize 2 SecondaryPos 2 PrimarySize 2 2 The following lines of code will then display the dialogue box as shown in Fig 56 and await a user response Display a dialog to select the stimulus display MonitorToUse uigetpref Monitors One Stimulus Monitor Selection Dialog Select which monitor to use 1 2 CheckboxString Always use this monitor switch MonitorToUse case 1 SelectedPos PrimaryLowerLeft PrimarySize case 2 SelectedPos SecondaryLowerLeft SecondarySize end Stimulus nix Select which monitor to use Always use this monitor EE JES
53. ewing distance screen resolution and scale The following sections describe the calibration process for each presentation device Dual VGA If you are using the Toolbox in a Windows dual display configuration you must calibrate the size of the stimulus display The Toolbox assumes a minimum of two monitors the Primary monitor is used to display Windows desktop and the Secondary monitor to display the visual stimulus It is the size and resolution of this Secondary stimulus monitor that needs to be calibrated Display Properties Themes Desktop Screen Saver Appearance Settings Drag the monitor icons to match the physical arrangement of your monitors ga Display 1 Plug and Play Monitor on VSG Visage Video Adaptor VP990 v Soreen resolution Color quality Less J More Highest 22 bit X 1024 by 768 pixels Identity Troubleshoot 7 Advanced Fig 5 Display Properties 16 Video Eyetracker Toolbox Calibration Before carrying out the calibration procedure ensure that the Windows desktop has been extended to both monitors and that the screen resolution and frame rates for the Primary Windows desktop and Secondary stimulus monitor agree with the minimum or recommended settings for both monitors as given in the Installation Manual If you intend to use videoEyetrace your Primary and Secondary monitors must have a resolution of 1024 x 768 pixels These setting
54. ge file must be on your MATLAB path ImageFile which Glasgow bmp Read the selected image file getting both the image and the palette data Image Palette imread ImageFile 7 m o Create a figure window in which to display the stimulus then stimulus on the figure window Load the bitmap image onto the selected monitor scaling appropriately FigurelHandle figure 1 AxesHandle axes set AxesHandle Position 00 1 1 oe oe Jo set FigurelHandle MenuBar none ToolBar none Position SelectedPos imagesc Image colormap Palette axis off Create Mimic Screen and load Bitmap image vetCreateMimicScreen vetSetMimicScreenDimensions 400 0 400 300 vetSetFixationPeriod 300 vetSetFixationRange 10 vetSetMimicPersistenceType CRS ptMotionAndFixations vetLoadBmpFileToMimicScreen ImageFile 1 Start tracking vetClearDataBuffer vetStartTracking oe Track for 30 seconds then stop tracking pause 30 vetStopTracking oe Tidy up by clearing away the camera and mimic Screens vetDestroyCameraScreen vetDestroyMimicScreen close figurelHandle oe Retrieve the recorded eye positions without removing them from the buffer Remove false DATA vetGetBufferedEyePositions Remove 67 Video Eyetracker Toolbox Communication 9 Display the retrieved positions figure 2 cla hold
55. ge of the mimic screen Width Specifies the width of the mimic screen Height Specifies the height of the mimic screen Example vetSetMimicScreenDimensions 2 10 200 150 will set the dimensions of the mimic screen on its parent window so that it is located 2 pixels from the left 10 pixels from the top is 200 pixels wide and 150 pixels high See also vetCreateMimicScreen vetDestroyMimicScreen vetMimicPersistence vetMimicPersistenceType vetLoadBitmapToMimicScreen vetMimicTraceColor vetSetMimicTraceColor The vetSetMimicTraceColor function specifies the colour of the trace on the mimic screen 113 Video Eyetracker Toolbox Appendices MATLAB Syntax vetSetMimicTraceColor mimicTraceColor Parameters mimicTraceColor A 3 by 1 or 1 by 3 RGB colour vector each component of which is a rational scalar in the interval 0 1 1 0 0 produces 100 red 0 1 0 produces 10096 green 0 0 1 produces 100 blue See also vetCreateMimicScreen vetDestroyMimicScreen vetSetMimicScreenDimensions vetSetMimicPersistence vetSetMimicPersistenceType vetLoadBitmapToMimicScreen vetSetStimulusDevice The vetSetStimulusDevice function allows you to select the stimulus device which will be used by the Video Eyetracker Toolbox for calibration purposes Valid devices are CRS deVSG i e a ViSaGe or VSG2 5 or CRS deVGA which is usually the Secondary Windows VGA Display Note to set the Toolbox to work with a custom
56. ical Specification 136 Appendix H Troubleshooting 137 System slow to calibrate 139 Video Eyetracker Toolbox Contents System fails to calibrate 139 System fails to track the eye 139 System fails to track the Pupil 140 System fails to track the Purkinje Glints 142 Eye tracking appears inaccurate 143 Eye tracking appears noisy erratic or intermittent 143 Notes 147 Video Eyetracker Toolbox Introduction d p In this section Overview of the Video Eyetracker Toolbox Family 10 CRS Eyetracker Toolbox 12 50 Hz Video Eyetracker Toolbox 12 250 Hz High Speed Video Eyetracker Toolbox 12 Images Acquired with 50 Hz and 250 Hz Systems 13 Video Eyetracker Toolbox Introduction Overview of Video Eyetracker Toolbox Family This manual is intended to cover the hardware and software use of the Cambridge Research Systems Video Eyetracker Toolbox family of eye trackers It is intended to be read only when all the steps in the corresponding Installation Manual have been successfully completed The Video Eyetracker Toolbox family of 50 Hz and 250 Hz video eye trackers share most of their functionality in common As a result this document will treat common functionality without reference to particular models and add sections specific to particular versions only where necessary The Video Eyetracker Toolbox allows you to add robust and accurate eye tracking to new or existing experimental systems In addition your Video
57. ing how long a fixation needs to last in order to be detected A sensible value for this parameter would be around 500ms accuracyLevel A scalar integer value drawn from the following set CRS acLow CRS acMedium CRS acHigh This value indicates how much movement will be tolerated when detecting a fixation imageFilename A char array Matlab string containing the filename and full path This parameter should be set to an empty string if targetShape has been set to anything other than tslmage Example vetCalibrateAdvanced 3 3 CRS tsSquare 0 5 1 0 0 0 0 1 90 100 CRS acMedium See also vetCalibrate vetCalibrationStatus vetSelectVideoSource vetStartTracking vetLoadCalibrationFile vetSaveCalibrationFile vetClearAllRegions This function deletes all Regions Of Interest from the Video Eyetracker Toolbox Regions of interest list See the help for vetAddRegion for a description of Regions Of Interest MATLAB Syntax vetClearAllRegions See also vetAddRegion vetDeleteRegion vetRegionCount 87 Video Eyetracker Toolbox Appendices vetClearDataBuffer The Video Eyetracker Toolbox maintains a buffer in which it records eye position and related eyetracking data If you are running multiple trials back to back and your experimental paradigm requires you to collect data separately for each trial this function will allow you to clear the buffer between trials MATLAB Syntax vetClearDataBuffer See
58. itle bar 37 Menu bar 38 Toolbar 38 Data windows 39 Status bar 40 Changing Stimuli and Settings 40 Changing View Options 41 Graph Tab 42 Configuration Tab 43 Mimic Tab 43 EIB Tab High Speed System only 44 Eye Tracking 45 Saving and Loading 48 Data Files ved 48 Custom Results Data File ved 49 MATLAB Data File mat 49 Mimic Image bmp 49 Communicating with the Toolbox 51 Synchronisation 50 Hz Systems 52 Synchronisation High Speed Systems 54 Eyetracker Interface Box Inputs 54 CRS ViSaGe 55 CRS VSG 2 5 56 Eyetracker Interface Box Outputs 57 Video Eyetracker Toolbox and MATLAB 57 Worked MATLAB Eye Tracking Examples 58 Example 1 Bitmap Stimulus with Dual VGA 58 Example 2 Triggered Saccade Stimulus for ViSaGe 68 Appendices 79 Appendix A MATLAB Commands 81 Functions 82 Events 83 vetAddRegion 84 vetCalibrate 85 vetCalibrateAdvanced 86 Video Eyetracker Toolbox Contents vetClearAllRegions 87 vetClearDataBuffer 88 vetClearMimicScreenBitmap 88 vetCreateCameraScreen 88 vetCreateMimicScreen 89 vetDeleteRegion 89 vetDestroyCameraScreen 90 vetDestroyMimicScreen 90 vetGetActiveRegion 90 vetGetBufferedEyePositions 91 vetGetCalibrated 93 vetGetEIB LEDsENABLED 93 vetGetEIB XYOutputType 93 vetGetFixationLocation 94 vetGetl atestEyePosition 95 vetGetMimicScreenBitmap 96 vetGetPupilCalibrated 96 vetGetRegionCount 97 vetGetResultsCount 97 vetGetTracking 98 vetGetToolboxVersion 98 vetGetVideoSourceFile 98 v
59. ix F for further details The Appearance sub window displays the current settings for all graphical traces which can be customised by clicking on the appropriate browse icons The EIB Trig ADC input trigger trace is 42 Video Eyetracker Toolbox videoEyetrace only available for the High Speed system See page 54 for further details Configuration Tab Video Eyetracker Options options Graph Configuration Mimic Window EIB Environment v Display warning before over writing unsaved data v Display NOT CALIBRATED warning on graph Record video stream to file Always track eye position Fixation Duration 300 Data file field separator character semi colon v Save Tracked field data as Yes No v Fig 35 Video Eyetracker Options Configuration Tab The Configuration tab displays a set of configuration settings covering visual warnings and the option to record the video stream to file In addition you can control the minimum duration necessary to identify a fixation and choose the type of field separator in the results data file The last option box controls whether the tracked status in each image is recorded in the results file as either an alphabetic Yes No or binary 1 0 format Mimic Tab The Mimic tab controls how the measured eye tracking trace is presented in the Mimic window The top dialog box controls the 43 Video Eyetracker Toolbox videoE
60. lation section of the High Speed Video Eyetracker Toolbox Installation Manual Trigger Cable to Power Cable VSG if required You must install Firewire Cable Firewire Cable software before lcs Kis aes connecting these N E Ts A Camera Cable BEE Gable Eyetracker Interface Box EIB Fig 49 Connections to Camera EIB and PC 54 Video Eyetracker Toolbox Communication CRS ViSaGe To receive digital timing triggers the High Speed system requires a trigger cable connecting from the Digital Trig port on your EIB see Fig 51 to the 5 pin Accessory connector on your ViSaGe see Fig 50 This frees up ViSaGe s 25 pin digital IO port for use with for example an EEG system Fig 50 Accessory Connector on ViSaGe The trigger signals are transmitted from the EIB via its USB port to the PC where it is logged in the DigitallO column of the videoEyetrace results file and the digitallO field in the MATLAB Data structure see Appendix G for further details However unlike the 50 Hz system the High Speed version does not output a digital tracking signal to the VSGs Digital Input line 7 but is instead output on the Tracked BNC connector see Fig 53 Fig 51 Eyetracker Interface Box Rear Panel The High Speed system can accept an input trigger from other external devices via the Trig In BNC connector on the EIB front panel see
61. lay screen Instruct the subject to fixate on the target The Toolbox will analyse all of the incoming images to identify a sequence of consistent pupil and Purkinje measurements Once the Toolbox is satisfied with the measurements it has collected the target will disappear and a new target will be displayed in a different location This procedure will repeat itself until an entire sequence of targets has been displayed to the subject Note The Toolbox will not proceed to the next target until it has identified a consistent cluster of measurements For this reason it is important that the subject remains as still as possible during calibration The size of a target is indicated underneath each target on the Scale Calibration Stick 27 Video Eyetracker Toolbox Calibration 4 After the subject has worked through the complete set of calibration targets the Toolbox will perform its internal calibration and present a second set of dots in the Calibration window see Fig 17 The dots show the screen target positions while the green lines show the relative positions where the system calculated the subject was actually looking A good calibration is indicated by a set of short green lines The following picture shows an example of a good calibration Subject Calibration subject calibration Pupil Target Diameter Save as Default Calibration Fig 17 Successful Calibration 5 Sometimes during the calibrati
62. lbox Communication Workspace a x a m NL se Base Name Value Class A11 1 0127 double A12 D 027078 double A13 0 08322 double A21 0 024665 double A22 1 0535 double A23 0 02163 double A31 D 0012532 double A32 D 00022519 double 3 DigitallO lt 11439x1 double double Dropped lt 11439x1 double double 7 E11 D 18003 double E12 2 9164e 005 double E21 5 5426e 005 double E22 0 19996 double E FicksLatitude lt 11439x1 double gt double H FicksLongitude 11439x1 double double FrameRate 50 double HelmholtzAzim 11439x1 double double HelmholtzEleva 11439x1 double double MeanX 158 89 double Mean Y 89 325 double lt ili lt Fig 64 Calibration and Tracked Eye Data as MATLAB Variables You can now use MATLAB to analyze plot or process any of your calibration or recorded eye results data The full code listing for demoVETsacc1 is as follows 74 Video Eyetracker Toolbox Communication function demoVETsaccl A simple VET demo to elicit reflexive horizontal saccades This demo requires a CRS ViSaGe global CRS crsLoadConstants oe oe oe oe Check we are running on a VSG ViSaGe CheckCard crsGetSystemAttribute CRS DEVICECLASS if CheckCard 7 error Sorry this demonstration requires a VSG ViSaGe end Select the VET camera to use vetSetStimulusDevice CRS deVSG errorCode vetSelectVideoSource CRS vsUserSelect if errorCode lt 0 e
63. lied with an example application called videoEyetrace which demonstrates the Toolbox s functionality in a ready to use form The videoEyetrace Window Start videoEyetrace by clicking on the videoEyetrace icon or via the Windows Start menu The main window provides a ready made user interface for the Toolbox see Fig 22 below The buttons on the menu bar and tool bar are used to control the type of stimulus displayed how the data is collected and how the test subject is calibrated toolbar pupil position untracked E Pupil diameter mm MI video window mimic window pupil diameter Fig 22 videoEyetrace Main Window 36 Video Eyetracker Toolbox videoEyetrace Initialising the Video Eyetracker Toolbox Camera videoEyetrace is designed to work with the whole Video Eyetracker Toolbox family as well as pre recorded video cam files Use the Select Video Source option from the File menu to reveal the following window select video source select source camera OK Video Eyetracker Camera ad Device Type Standard Camera Standard Camera Standard Camera l High speed Camera High speed Camera High speed Camera High speed Camera High speed Camera Fig 23 Select Video Source Window The EyeLock icon allows you to select between the different cameras and their range of sampling frequencies If you are running a 50 Hz system and your camera was shipped with a Camera l label select Standard Camera
64. ll times and refocus if necessary The eyelid may be obscuring the pupil slightly which would reduce the accuracy of finding the pupil centre e The system may have been poorly calibrated or lost calibration due to movement of one of the headrest components Ensure that all headrest adjustments are correctly and securely set and then recalibrate Eye tracking appears noisy erratic or intermittent In normal use the Toolbox will track the subject s point of fixation with angular noise of between one quarter and one half of a degree which is approximately 2 5 5mm positional error at a viewing distance of 500mm These specifications can be achieved by ensuring that The radial lines on the iris are clearly visible i e the pupil iris is in focus not the eyelashes There are no bright spurious glints close to the pupil e The pupil is in the centre of the image and not occluded by the eyelid in any way These are illustrated in Fig 80 143 Video Eyetracker Toolbox Appendices Fig 80 Good Video Imaging Conditions High Speed Eyetracker Only Intermittent tracking can also be caused by the PC failing to keep up with the video frame rate If this occurs the user should check that the PC is not performing any non vital background processes If the problem continues the user should either reduce camera speed in the Select Video Source window or if this is not viable it may be necessary to consider
65. lts in the picture always being stretched to fill the entire mimic screen See also vetCreateMimicScreen vetDestroyMimicScreen vetSetMimicScreenDimensions vetMimicPersistence vetMimicPersistenceType vetMimicTraceColor vetSetDeviceParameters vetSaveCalibrationFile This function stores a subject s calibration data to a file specified by Filename for future use For example this file could be used to provide a default calibration to be used with subjects that cannot fixate or are unable to calibrate for other reasons MATLAB Syntax vetSaveCalibrationFile filename 102 Video Eyetracker Toolbox Appendices Parameters Filename This specifies the name of the file which the calibration data will be saved to NB Windows standard separation characters single backslashes should be used E g C myDirectory myCalibFile scf See also vetCalibrate vetCalibrateAdvanced vetLoadCalibrationFile vetSaveMimicScreenBitmap Use this function to save a background bitmap copy of the mimic screen to a Windows bitmap image file MATLAB Syntax vetSaveMimicScreenBitmap Filename Parameters Filename This specifies the name of the file which the mimic window bitmap will be saved to This parameter should be a MATLAB char array string including the full path to the image file See also vetCreateMimicScreen vetGetMimicScreenBitmap vetLoadBitmapToMimicScreen vetSaveResults The Video Eyetracker Tool
66. mination sources will appear slightly larger in the images obtained by the High Speed version as more illumination is required to obtain clear high contrast images Fig 4 Comparison of 50 Hz left and 250 Hz right Images 13 Video Eyetracker Toolbox Calibration d In this section Monitor Configuration and Calibration 16 Dual VGA 16 Dual VGA with GazeTracker 18 CRS Visual Stimulus Generator 19 Non Computer based Custom Display 20 Positioning and Focusing your Subject 21 Subject Calibration 24 Pupil Scale Calibration 25 Subject Gaze Calibration 27 Hardware states after Subject Calibration 33 CRS Visual Stimulus Generator 33 Dual VGA 34 Custom Device 34 15 Video Eyetracker Toolbox Calibration Monitor Configuration and Calibration The Video Eyetracker Toolbox is primarily designed to work in a dual computer screen configuration with separate displays for the Windows desktop and visual stimulus It is also possible to use the system with a non computer based display such as a TV blackboard or LED array However in this scenario you must manually configure and calibrate your display source and subject To provide a second stimulus display you can either use a single PC graphics card with dual outputs two separate PC graphics cards or a CRS Visual Stimulus Generator e g the ViSaGe Whichever solution you choose to use it is extremely important to correctly calibrate the stimulus display in terms of vi
67. mm holes through the remainder of the worksurface A cross section through one of these holes is shown below 20 mm diameter 4 20 mm depth table thickness 30 mm min gt 8 1mm diameter Fig 70 EyeLock Hole Cross Section To prevent unnecessary stressing of the headrest it is important that these holes are positioned accurately relative to each other and to the edge of the workbench Use the template below to ensure all holes are drilled correctly Take care not to damage this template so that it remains available for future use Base of EyeLock Headrest va 85 mm diameter 200 mms 20 mm diameter hole 8 1 mm diameter hole for vertical supports for M8 bolt Fig 71 Positioning of EyeLock Holes 128 Video Eyetracker Toolbox Appendices d d Appendix D Technical Specification The following specifications state what the Toolbox is designed to achieve in normal use provided reasonable care has been taken during imaging set up and calibration The stated variation in tracking accuracy is dependent on each subject s ability to fixate on the calibration targets 129 Video Eyetracker Toolbox Technical Specifications Measurement Technique Imaging Technique Sampling Frequency 50 Hz High Speed Latency Resolution Pupilometry Range Pupilometry Accuracy Tracking Accuracy Horizontal Range Vertical Range Allowable Head Movement Measurement Units Appendice
68. n the subject s direction 90 Video Eyetracker Toolbox Appendices of gaze enters a region the MATLAB command specified with the vetSetCallbackRegionChanged function is executed MATLAB Syntax activeRegion vetGetActiveRegion Return Values activeRegion An integer scalar 1 by 1 matrix in the interval 1 int_max showing the current active region See also vetAddRegion vetGetLatestEyePosition vetGetBufferedEyePositions This function returns eye position data from the buffer that is created and maintained by the Video Eyetracker Toolbox software Eye positions that are returned by this function are by default removed from the data buffer so subsequent calls will not return duplicate data If there are no points available in the buffer a structure filled with zeros is returned together with a non zero 1 ErrorCode NOTE In order to determine if the subject is fixating eye position data is delayed by the user specifiable fixation period defined using the vetSetFixationPeriod function This function has two optional parameters a Boolean logical value removeFromBuffer and a numeric integer value maximumPoints If the removeFromBuffer parameter is set to False then data points are left in the VET buffer allowing them to be retrieved again by a second call If the removeFromBuffer parameter is set to True then data points are deleted from the buffer after they are read The maximumPoints parameter allo
69. n your PC The original Video Eyetracker Toolbox 50 Hz fixation monitor has two camera types CRS vsCamera refers to the progressive scan camera type and CRS vsCameral refers to the newer mode Camera Interlaced mode Cameras can be identified by a white or yellow Camera l sticker on the casing Filename If the selected video source is CRS vsFile then this parameter should contain a full path to a prerecorded CAM file e g C myDirectory myVideo cam If the selected video source is CRS vsUserSelect then this parameter should contain a CAM file to default to If the selected video source is a camera of some description then this parameter should be an empty string 105 Video Eyetracker Toolbox Appendices See also vetStartRecordingToFile vetStopRecording vetSetStimulusDevice vetVideoSourceFile vetSetCameraScreenDimensions This function is used to set the dimensions of the camera screen control Note that the camera screen control does not have to exist before this function is called MATLAB Syntax vetSetCameraScreenDimensions Left Top Width Height Parameters Left Specifies the left edge of the camera screen Top Specifies the top edge of the camera screen Width Specifies the width of the Camera screen Height Specifies the height of the Camera screen Example SetCameraScreenDimensions 10 20 400 300 will set the dimensions of the camera screen on its parent window so that it is l
70. ned in the trigIn field and are in the range of 1 This input also appears in videoEyetrace as the EIB Trig ADC graphical plot and is also logged as part of the ved results data file VSG2 5 Feature p Digital Connector Fig 52 25 pin Digital I O on VSG2 5 56 Video Eyetracker Toolbox Communication Eyetracker Interface Box Outputs Fig 53 Eyetracker Interface Box Front Panel Apart from the Trig IN BNC the other BNC connectors output the recorded eye position data which can then be logged using external equipment The Pupil output is statically scaled at 0 1 volts per mm of pupil diameter Trig Out replicates any signal fed into Trig In and Tracked outputs a standard TTL signal of either 0 or approximately 5 volts The scales of the X and Y outputs can be set using the vetSetEIB_XYOutputType command which is detailed in Appendix A or within videoEyetrace under the EIB tab of Graph Options Video Eyetracker Toolbox and MATLAB MATLAB is a data analysis and visualisation application built around a simple scripting language and is widely used in the creation of models and in the analysis of experimental results By providing tools to help the vision researcher control and analyse their experiments from a single environment we hope to reduce the training burden on new entrants to the field as well as making it easier for existing experimenters to integrate their experimental design and control with their
71. ocated 10 pixels from the left 20 pixels from the top is 400 pixels wide and 300 pixels high See also vetCreateCameraScreen vetDestroyCameraScreen vetSetDeviceParameters If you are using a custom stimulus device such as an LED array or laser galvanometer system then you should use this function to provide the Video Eyetracker Toolbox with information about your stimulus device such as viewing distance and screen size 106 Video Eyetracker Toolbox Appendices This information is used during the VET calibration procedure to specify target locations for your calibration routine If you are using a Cambridge Research Systems VSG2 5 ViSaGe or a secondary Windows VGA display you should use vetSetStimulusDevice instead of vetSetDeviceParameters as viewing distance and stimulus display size are obtained automatically from the VSG Software Library or VGA Calibration utility MATLAB Syntax vetSetDeviceParameters device ViewingDistance_mm Width_mm Height_mm Parameters Device Only use the following value CRS deUser Note Setting device to deUser has a direct effect on the function vetLoadBitmapToMimicwindow ViewingDistance_mm Specifies the viewing distance of the subject to the stimulus display screen in millimetres Width_mm this parameter specifies the width of the display screen in millimetres If Device is not set to deUser this parameter is ignored Height mm this parameter specifies the Height of
72. on plot DATA fickPositions 1 b Longitude in blue hold on plot DATA fickPositions 2 r Latitude in red grid on Save the results to disk as a comma delimited file CurrentDirectory cd tempfile CurrentDirectory myResults csv vetSaveResults tempfile CRS ffCommaDelimitedNumeric Clear the data buffer vetClearDataBuffer Example 2 Triggered Saccade Stimulus for ViSaGe Example 2 is specifically designed to demonstrate how the High Speed system can be integrated with a Cambridge Research Systems ViSaGe for a typical oculomotor research application This example calibrates a test subject generates a series of fixation and saccade targets and tracks the subject s eye positions The eye movement results and stimulus timing triggers are then displayed as a graph and saved to a file The first section of code initialises the ViSaGe selects the video source creates Camera and Mimic windows and performs calibration as in the previous MATLAB example function demoVETsaccl A simple VET demo to elicit reflexive horizontal saccades This demo requires a ViSaGe global CRS crsLoadConstants Check we are running on a VSG ViSaGe CheckCard crsGetSystemAttribute CRS DEVICECLASS if CheckCard 7 error Sorry this demonstration requires a VSG ViSaGe end Select the VET camera to use vetSetStimulusDevice CRS deVSG errorCode vetSelectVideoSource CRS vsUserSelect if errorCode lt 0
73. on procedure the subject may focus on the wrong part of the display screen away from the target If this happens you will see a large green line or a cluster of large green lines in one region of the Calibration window You may also see smaller lines from surrounding targets being pulled in the opposite direction of the larger lines See figure Fig 18 If this happens it is possible to select the individual targets that contain the largest errors for recalibration To select the targets you want to represent just left click on them with your mouse The calibration procedure can now be repeated on these selected targets by clicking the Start button as before Note left clicking on a selected target will deselect it 28 Video Eyetracker Toolbox Calibration Subject Calibration Gy subject calibration Pupil Target Diameter 2 mm Apply Save as Default Calibration Fig 18 Calibration window illustrating two bad calibration targets that have been selected Subject Calibration subject calibration Pupil Target Diameter 2 mm Apply C Save as Default Calibration Fig 19 Calibration window illustrating the same calibration as above after the two targets have been re calibrated 29 Video Eyetracker Toolbox Calibration Sometimes when calibrating a subject it may be necessary to change the Toolbox s default settings You may want to make it easier harder for the subject to fixate on a target or try
74. options Front Panel LEDs v Enable Front Panel LED Operation X Y BNC Outputs Output Type mm 0 01 mm Fig 37 Video Eyetracker Options EIB Tab Eye Tracking Once the calibration procedure is complete and a stimulus has been loaded press the Run icon on the Toolbar to enter tracking mode Fig 38 Run Icon 45 Video Eyetracker Toolbox videoEyetrace This will initiate tracking and output the results in the graph window in real time A typical graph output for a sinusoidal smooth pursuit task is shown in Fig 39 Fig 39 Smooth Pursuit Task The horizontal axis represents time in seconds while the vertical axis represents eye gaze screen position in millimetres In this example the target was moving horizontally which is indicated here by the sinusoidal red horizontal trace while the vertical blue trace represents biological and system noise A NS bd 4 F AB Fig 40 Graph Zoom Icons The horizontal and vertical scales of the output graph can be changed before during or after tracking using the above icons available on the main toolbar During tracking videoEyetrace will also project the recent eye positions into the mimic window and will detect and display any fixations as circles with radii proportional to their duration A typical example of this is shown in Fig 41 46 Video Eyetracker Toolbox videoEyetrace Fig 41 Mimic window showing locus of tracked eye positions and fixations
75. orCode This may happen if you attempt to read data soon after tracking starts as eye position data can be delayed by the user specifiable fixation period defined using the vetSetFixationPeriod command MATLAB Syntax Data vetGetLatestEyePosition Return Values Data A MATLAB structure containing the VET eye position data Data timeStamps Time in milliseconds since tracking started Data tracked A Boolean value where 0 FALSE and 1 TRUE Data calibrated A Boolean value where 0 FALSE and 1 TRUE Data pupilDiameter Pupil diameter in millimetres Data mmPositions Direction of gaze projected onto stimulus display in mm X Y Data ficksPositions Eye rotation in Fick coordinates Longitude Latitude Data helmholtzPositions Eye rotation in Helmholtz coordinates Elevation Azimuth Data region Current region number Data fixation A Boolean value where 0 FALSE and 1 TRUE 95 Video Eyetracker Toolbox Appendices Data digitallO A value containing the combined digital IO values see page 53 for details Data dropped A Boolean value where 0 FALSE and 1 TRUE Data trigln High Speed system only EIB Trigger input values See also vetlsEyeDataAvailable vetGetBufferedEyePosition vetGetMimicScreenBitmap This function returns the mimic window bitmap as a MATLAB matrix MATLAB Syntax ImageMatrix vetGetMimicScreenBitmap Return Values ImageMatrix A MATLA
76. ow as shown in Fig 57 Track for 30 seconds then stop tracking vetClearDataBuffer vetStartTracking pause 30 vetStopTracking Tidy up by clearing away camera and mimic screens vetDestroyCameraScreen vetDestroyMimicScreen close figurelHandle 62 Video Eyetracker Toolbox Communication VET Mimic Window VET Mimic Window Fig 57 Mimic Window with Motion and Fixations Trace The tracked results will now be plotted in Fick angular coordinates with longitude in blue and latitude in red Some example results are shown in Fig 58 Retrieve the recorded eye positions without removing them from the buffer Remove false DATA vetGetBufferedEyePositions Remove Display the retrieved positions figure 2 cla holdon plot DATA fickPositions 1 b Longitude in blue hold on plot DATA fickPositions 2 r Latitude in red grid on 63 Video Eyetracker Toolbox Communication Figure 3 DER File Edit View Insert Tools Desktop Window Help OB so Fig 58 Tracked Eye Data in Fick Angular Co ordinates Finally the results are saved to disk and the eye position data buffer is then deleted Save the results to disk as a comma delimited file CurrentDirectory cd tempfile CurrentDirectory myResults csv vetSaveResults tempfile CRS ffCommaDelimitedNumeric Clear the data buffer vetClear
77. ox Notes 149 Video Eyetracker Toolbox Notes 150 Video Eyetracker Toolbox Notes 151 Video Eyetracker Toolbox Notes 152
78. pending on whether the Video Eyetracker Toolbox is currently tracking MATLAB Syntax tracking vetGetTracking Return Values tracking A boolean value 1 by 1 matrix where 0 FALSE and 1 TRUE See also vetStartTracking vetStopTracking vetGetToolboxVersion This function returns the current video eyetracker toolbox version number MATLAB Syntax toolboxVersion vetGetToolboxVersion Return Values toolboxVersion The Video Eyetracker Toolbox Software version number vetGetVideoSourceFile If the Video Eyetracker Toolbox is set to play and track a pre recorded video file then calling vetGetVideoSourceFile will return the full name and file path of the CAM file that has been selected MATLAB Syntax videoSourceFile vetGetVideoSourceFile 98 Video Eyetracker Toolbox Appendices Return Values videoSourceFile A char array a Matlab string between 1 and 500 characters long inclusive of the terminating null character See also vetSelectVideoSource vetGetVideoSourceType The Video Eyetracker Toolbox can obtain a video stream from a number of different camera types The function vetSelectVideoSource allows you to specify the video source that you are using This function returns the value that has been set using vetSelectVideoSource or zero if no source has yet been selected MATLAB Syntax videoSourceType vetGetVideoSourceType Return Values videoSourceType An intege
79. ply equally to the 50 Hz and 250 Hz versions Before proceeding further it may be useful to make a brief comparison between the video images obtained from the two systems Fig 74 50 Hz Left and 250 Hz Right Video Images These images of the same test subject were taken using the 50 Hz CCD and 250 Hz CMOS cameras The noticeable differences are clarity scale and the size of the Purkinje glints The lower clarity in the 250 Hz system is partly due to using a high speed CMOS sensor and partly due to reduced depth of focus from using a wider aperture lens to gather sufficient light However the effects of lower image clarity are compensated for by using higher optical magnification and larger Purkinje glints In tests both systems meet the same specifications for accuracy although the High Speed version will tolerate slightly less head movement 138 Video Eyetracker Toolbox Appendices System slow to calibrate The calibration process requires the subject to fixate on a series of screen targets The system will not advance to the next screen target until it has made a number of consistent measurements from the current fixation A delay in advancing to the next screen target may be caused by head movement which is tolerated during normal use but not during calibration Alternatively slow calibration may be the result of intermittent or noisy eye tracking see relevant sections below System fails to calibrate Refusal to ad
80. procedure as described in the previous section for a standard Dual VGA configuration Since GazeTracker uses the Primary Windows monitor to display stimuli you may find it easier to use if you drag the desktop icons and Windows Task Bar onto the Secondary monitor s desktop 18 Video Eyetracker Toolbox Calibration CRS Visual Stimulus Generator If you are using the Toolbox with a CRS Visual Stimulus Generator e g ViSaGe or VSG2 5 you must configure the stimulus monitor using the VSG Software Library First open vsgDesktop from the Start menu via Start Programs VSG Software Library VSG Desktop and click on the Monitor Configuration plug in A red and green chessboard will be presented on the VSG stimulus display Monitor Configuration c ato Manual Curent Settings D Device Moritor Flame Rate Scan Rate Width Height VSG ViSaGe 256MB VSG 71 01 0040 SONY GDM F520 100 00 8140 1024 768 TesiPatten Piel Sie Calbration E jer fr the VSG to prod sized objects on screen Fist setup with ofthe ma bec the equi the snall square tuely square Size of Square 00 v pie Height of Square 37 40 mn j Default Viewing Distance 480 Fig 8 Monitor Configuration within vsgDesktop Now click on the Calibration tab located in the main window of vsgDesktop Th
81. r Toolbox Communication Synchronisation 50 Hz Systems The cable supplied with your Video Eyetracker Toolbox terminates with a 9 pin D type connector and a 25 pin D type connector The 9 pin connector attaches to the Picolo frame grabber and the 25 pin connector to a Digital I O port on either a VSG2 5 Feature Connector or a ViSaGe as shown in Fig 45 VSG2 5 Feature Digital I O Connector Fig 45 25 pin Digital I O on ViSaGe Left and VSG2 5 Right Note It is not necessary to connect the 25 pin D type end of the cable unless you are planning to synchronise the data collected by the Toolbox with your stimulus presentation system Trigger Cable to VSG if required Cable to Picolo Frame grabber Fig 46 Cable Connections for Triggering 52 Video Eyetracker Toolbox Communication The Toolbox s video capture card has four digital I O lines The first three least significant of these lines are set as input lines and their states are read every frame Thus it is possible to feed in three independent signals that will be read by the Toolbox every frame In order to guarantee that your trigger values are read by the Toolbox they should last for at least a frame length as the Toolbox only samples the lines once every frame If you are using a VSG it is advisable to create stepped rather than pulsed triggers which are only modified when the stimulus changes See
82. r scalar value one of the following constants CRS vsCamera Progressive scan 50 Hz camera no identifying label on camera case CRS vsFile Takes video from a pre recorded saved CAM file CRS vsCameral Interlaced mode 50 Hz camera white Camera l label on camera case CRS vsHighSpeedCamera60 High Speed Camera running at 60Hz CRS vsHighSpeedCamera100 High Speed Camera running at 100Hz CRS vsHighSpeedCamera160 High Speed Camera running at 160Hz CRS vsHighSpeedCamera200 High Speed Camera running at 200Hz CRS vsHighSpeedCamera250 High Speed Camera running at 250 Hz 99 Video Eyetracker Toolbox Appendices See also vetSelectVideoSource 6 1 6 This function returns the value TRUE 1 if there are any results in the data buffer Results can only be entered into the buffer if the Toolbox has been successfully calibrated and tracking has been started MATLAB Syntax eyeDataAvailability vetIsEyeDataAvailable Return Values eyeDataAvailability An integer scalar 1 by 1 matrix holding the value 1 if eye data is available and 0 if it is not See also vetClearDataBuffer vetLatestGetEyePosition vetLoadBmpFileToMimicScreen Use this function to load a background bitmap onto the mimic screen The mimic screen is a graphical representation of eye location If a scaled image of th
83. ramming tool that supports Microsoft COM The CRS Toolbox for MATLAB contains the commands you need to use to control the Video Eyetracker Toolbox To download the CRS Toolbox for MATLAB login to the CRS Support Portal www crsltd com support login What follows is a listing of the functions that make up the MATLAB interface and the ways in which they can be used A description of the COM programming interface and additional demonstrations are available from the Cambridge Research Systems support portal www crsltd com support login 81 Video Eyetracker Toolbox Appendices vetAddRegion 84 vetCalibrate 85 vetCalibrateAdvanced 86 vetClearAllRegions 87 vetClearDataBuffer 88 vetClearMimicScreenBitmap 88 vetCreateCameraScreen 88 vetCreateMimicScreen 89 vetDeleteRegion 89 vetDestroyCameraScreen 90 vetDestroyMimicScreen 90 vetGetActiveRegion 90 vetGetBufferedEyePositions 91 vetGetCalibrated 93 vetGetEIBLEDsEnabled 93 vetGetEIB_XYOutputType 93 vetGetFixationLocation 94 vetGetLatestEyePosition 95 vetGetMimicScreenBitmap 96 vetGetPupilCalibrated 96 vetGetRegionCount 97 vetGetResultsCount 97 vetGetTracking 98 vetGetToolboxVersion 98 vetGetVideoSourceFile 98 vetGetVideoSourceType 99 vetlsEyeDataAvailable 100 vetLoadBmpFileToMimicScreen 100 vetLoadCalibrationFile 101 vetLoadBitmapToMimicScreen 102 vetSaveCalibrationFile 102 vetSaveMimicScreenBitmap 103 vetSaveResults 103 vetSelectVideoSource 104 vetSetCameraScreenDimensions
84. royMimicScreen vetSetMimicScreenDimensions vetSetMimicPersistence vetLoadBitmapToMimicScreen vetSetMimicTraceColor vetSetMimicPersistenceType vetSetMimicPersistenceType vetSetMimicPersistenceType specifies the type of trace that is displayed in the mimic window control To display a continuous trace illustrating eye position vetSetMimicPersistenceType should be set to CRS ptMotion To display eye position displayed as growing Fixation circles linked by motion traces vetSetMimicPersistenceType should be set to CRS ptMotionAndFixation Finally to display just growing Fixations set vetSetMimicPersistenceType to CRS ptFixations MATLAB Syntax vetSetMimicPersistenceType mimicPersistenceType Parameters mimicPersistenceType An integer value 1 by 1 matrix which should be one of the following constants CRS ptMotion CRS ptFixations CRS ptMotionAndFixations 112 Video Eyetracker Toolbox Appendices See also vetCreateMimicScreen vetDestroyMimicScreen vetSetMimicScreenDimensions vetSetMimicPersistence vetLoadBitmapToMimicScreen vetSetMimicTraceColor vetSetMimicPersistenceT ype vetSetMimicScreenDimensions Use this function to set the dimensions of the mimic screen Note that this function should be called before the mimic screen is created MATLAB Syntax vetSetMimicScreenDimensions Left Top Width Height Parameters Left Specifies the left edge of the mimic screen Top Specifies the top ed
85. rror Video Source not selected end Calibrate the video eyetracker vetCreateCameraScreen errorCode vetCalibrate if errorCode 0 error Calibration not completed end Create the Mimic Screen window vetCreateMimicScreen vetSetMimicScreenDimensions 400 0 400 300 vetSetFixationPeriod 300 vetSetFixationRange 10 vetSetMimicPersistenceType CRS ptMotionAndFixations Set the palette and pen levels for drawing red and green dots on a mean grey background GreyLevel 1 GreyRGB 0 5 0 5 0 5 crsPaletteSetPixelLevel GreyLevel GreyRGB RedLevel 2 RedRGB 1 0 0 0 0 0 crsPaletteSetPixelLevel RedLevel RedRGB GreenLevel 3 GreenRGB 0 0 1 0 0 0 crsPaletteSetPixelLevel GreenLevel GreenRGB oe Draw the fixation point and target on their respective video pages whilst displaying a blank page to hide the drawing TargetSize crsDegreesToPixels 0 3 0 3 BlankPage 1 crsClearPage BlankPage GreyLevel FixationPage 2 crsClearPage FixationPage GreyLevel oe oe 75 Video Eyetracker Toolbox Communication TargetPage 3 crsClearPage TargetPage GreyLevel crsSetDisplayPage BlankPage Draw a green fixation point on the fixation page crsSetDrawPage FixationPage crsSetPenl GreenLevel crsDrawOval 0 0 TargetSize Draw a red target on the target page crsSetDrawPage TargetPage crsSetPenl RedLevel crsDrawOval 0 0 TargetSi
86. s Video utilising Pupil and dual first Purkinje images Infrared camera plus 2 infrared sources 50 Hertz Up to 250 Hertz 20 msec 4 100 msec 0 05 Degree 3 10 Millimetres 0 5 Millimetres 0 125 0 25 Degree 40 Degrees 4 20 Degrees 12 Millimetres 50 Hz 8 Millimetres 250 Hz Fick and Helmholtz co ordinates in degrees or screen position in millimetres The maximum 100 msec latency figure quoted for the High Speed System is due to the time required for the Eyetracker Interface Box to create an analogue output for the tracked eye positions 130 Video Eyetracker Toolbox Appendices p AppendixE Hot Mirror Spectral Transmission 131 Video Eyetracker Toolbox Appendices EN N a e e e e e o o o 2 o eo o e e e E 22 E eas x so a Ti 3 a 8 zs 00001 0 00 1 Fig 72 Spectral Transmission of Infrared Mirror 132 Video Eyetracker Toolbox Appendices p 133 Video Eyetracker Toolbox Appendices Technical Description The Video Eyetracker Toolbox is able to return the view position either in horizontal vertical millimetres or in Fick or Helmholtz angle co ordinates The Fick system of angular coordinates specifies the horizontal longitude component first followed by the vertical latitude component The Helmholtz system specifies the vertical elevation component first followed by th
87. s 100 red 0 1 0 produces 100 green 0 0 1 produces 100 blue See also vetCreateMimicWindow vetSetMimicScreenDimensions vetSetMimicBackgroundColor vetSetMimicTraceColor vetSetMimicPersistence The vetSetMimicPersistence function specifies the duration of the trace that appears on the mimic screen For example setting mimicPersistence to 2 will cause the mimic screen control to display eye positions for the most recent 2 seconds only MATLAB Syntax vetSetMimicPersistence mimicPersistence Parameters mimicPersistence A positive rational scalar 1 by 1 matrix greater than or equal to 0 05 This value has units of seconds See also vetCreateMimicScreen vetDestroyMimicScreen vetSetMimicScreenDimensions vetSetMimicPersistenceType vetLoadBitmapToMimicScreen vetSetMimicTraceColor vetSetMimicPersistenceStyle The vetMimicPersistenceStyle function specifies the appearance of the mimic trace Setting vetMimicPersistenceType to CRS psConstant will cause the trace to appear as a constant color good for picture search 111 Video Eyetracker Toolbox Appendices tasks If it is set to psFade the trace will gradually fade into the background MATLAB Syntax vetSetMimicPersistenceStyle mimicPersistenceStyle Parameters mimicPersistenceStyle An integer value 1 by 1 matrix which should be one of the following constants CRS psConstant CRS psFade See also vetCreateMimicScreen vetDest
88. s at once using the vetClearAllRegions function MATLAB Syntax vetDeleteRegion RegionIndex See also vetAddRegion vetClearAllRegions vetRegionCount 89 Video Eyetracker Toolbox Appendices vetDestroyCameraScreen This function destroys the camera screen control created using the vetCreateCameraScreen function MATLAB Syntax vetDestroyCameraScreen See also vetCreateCameraScreen vetSetCameraScreenDimensions vetDestroyMimicScreen This function destroys the mimic screen A mimic screen can be created using the vetCreateMimicScreen function Only one mimic screen can exist at any one time Mimic screens are used to display the recent gaze direction in a graphical format MATLAB Syntax vetDestroyMimicScreen See also vetCreateMimicScreen vetSetMimicScreenDimensions vetMimicPersistence vetMimicPersistenceType vetLoadBitmapToMimicScreen vetMimicTraceColor vetGetActiveRegion This function returns the current active Video Eyetracker Toolbox Region Of Interest This is the region that the subject s direction of gaze currently intersects If the subject s direction of gaze does not currently intersect a specified Region Of Interest this function will return a negative number 1 Regions of interest can be specified using the vetAddRegion function and are intended to assist with fixation monitoring tasks The current region is also returned by vetGetLatestEyePosition and vetGetBufferedEyePositions Whe
89. s can be configured by a right mouse click on the desktop and choosing the Properties option to open the window shown in Fig 5 Once the Primary and Secondary screens have been suitably configured start the Toolbox s VGA Calibration application located in the Video Eyetracker Toolbox directory or from the desktop via Start Programs Video Eyetracker Toolbox VGA Calibration The VGA Calibration application will create a white square on a black background on the Secondary monitor and open the following window on the Primary monitor VGA Pixel Size Calibration Gy vga calibration The system needs to know how big the pixels on the display are in order to calibrate sizes correctly To correctly calibrate your display please adjust your monitors width and height settings as required to make the white square truly square Use a ruler to measure the height of the box on the second display and then enter the value into the edit box below Click OK when finished to save the value Size of Square 24 0 a mm Advanced Cancel i Fig 6 VGA Pixel Size Calibration To calibrate the Secondary monitor you must measure the width and height of the square using a ruler and if necessary adjust the horizontal or vertical scales on the display monitor itself to ensure that both measurements are equal When the white stimulus is truly square enter its height width in the Size of Square box and click OK Note this VGA Calibration pro
90. tRecordingToFile to record the currently selected source to a file The Toolbox does not perform any form of compression on the recorded stream and consequently requires a lot of space on a drive to record a file 30 seconds of recorded data at 50 Hz requires approximately 200 Mb of storage MATLAB Syntax vetStartRecordingtoFile filename Parameters Filename A char array Matlab string containing a full filename and path This specifies the name of the file which the streamed video signal will be saved to You must specify the full path to your CAM file C myDirectory myVideo cam See also vetSelectVideoSource vetStopRecording vetStartTracking This function initiates tracking and the collection of data Note that any data that has already been collected will be lost when this call is made and data will only be collected if the subject has been successfully calibrated You can however load a default calibration file instead if your subject is unable to fixate or difficult to calibrate for other reasons MATLAB Syntax vetStartTracking See also vetStopTracking 116 Video Eyetracker Toolbox Appendices vetStopRecording Call vetStopRecording to stop the Toolbox recording the current video stream to file MATLAB Syntax vetStopRecording See also vetStartRecording vetSelectVideoSource vetStopTracking This function instructs the Video Eyetracker Toolbox to stop tracking Eye position data will r
91. the File menu can be used to deselect any of the default file entries shown in Fig 43 It also provides an option to modify the field separator character and include a time field if required This field is expressed in milliseconds and starts at zero for the first results data entry Temporal quantisation is dependant on the sampling rate of your Selected Video Source Save Custom Data File File name Include column headings Fields C Time v Tracked Yes No v v Horizontal Position v Vertical Position v Pupil Diameter v Digital 1 0 Data file field separator character semi colon v Cancel Fig 44 Custom Data File Options Window MATLAB Data File mat The results data file can also be saved as a MATLAB mat format from the File menu Mimic Image bmp The Mimic image window complete with mimic trace can be saved as a Bitmap image from the File menu 49 Video Eyetracker Toolbox videoEyetrace 50 Video Eyetracker Toolbox Communication d Communicating with the Toolbox In this section Synchronisation 50 Hz Systems 52 Synchronisation High Speed Systems 54 Eyetracker Interface Box Inputs 54 CRS ViSaGe 55 CRS VSG2 5 56 Eyetracker Interface Box Outputs 57 Video Eyetracker Toolbox and MATLAB 57 Worked MATLAB Eye Tracking Examples 58 Example 1 Bitmap Stimulus with Dual VGA 58 Example 2 Triggered Saccade Stimulus with VSG 68 51 Video Eyetracke
92. the Toolbox will track and measure it Successful tracking is indicated by a green outline located around the circumference of the target The current estimate for the size of the target will be displayed in the top right corner of the camera window This value is calculated based upon the current scaling factor 4 Move the Scale Stick so that it is focused on the number shown in the Pupil Target Diameter box 8mm by default If you want to set the scale using a different size calibration dot choose a different size from the drop down box labelled Pupil 26 Video Eyetracker Toolbox Calibration Target Diameter located on the Calibration window The value selected in this box should always be set to the same size as the target being tracked in the camera window Pupil Target Diameter 3 mm Fig 16 Pupil Target Diameter Dropdown Box Now click the Apply button to set the new scaling factor The reported diameter of the target will now be identical or very close to the diameter of the target displayed in the camera window Measuring the remaining targets on the calibration stick can be used as a method to check the accuracy of the calibration over a large range 6 Click on the Scale button again to exit the pupil scale mode Subject Gaze Calibration 1 With the subject positioned correctly in the EyeLock headrest click the Start button on the Calibration window A calibration target will appear on the stimulus disp
93. the display screen in millimetres If Device is not set to deUser this parameter is ignored Example vetSetDeviceParameters CRS deUser viewingDistance mm width mm height mm See also vetSetStimulusDevice vetSetViewingDistanceMM vetLoadBitmapToMimicWindow vetSetEIB LEDsENABLED This property sets the Eyetracker Interface Box LEDs on or off Set this property to on true in order to turn the Eyetracker 107 Video Eyetracker Toolbox Appendices Interface Box LEDs on and set this property to off false in order to turn the LEDs off MATLAB Syntax vetSetEIB_LEDsEnabled LEDsEnabled See also vetVideoSourceType vetGetEIB_LEDsEnabled vetSetEIB XYOutputType This property allows you to select the XY analog output type of the Eyetracker Interface Box The values output to the two analog channels These can either be 100 mm per volt 1000 mm per volt degrees elevation azimuth Helmholtz or degrees longitude latitude Fick This value must be one of the following constants CRS otMM100 MM 100 mm V CRS otMM1000 MM 1000 mm V CRS otHelm1 0 1 Volts per degree Helmholtz coordinates CRS otFicks1 0 1 Volts per degree Fick coordinates MATLAB Syntax vetSetEIB XYOutputType XYOutputType See also vetGetLatestEyePosition vetVideoSourceType vetGetEIB XYOutputType vetSetFixationPeriod Set the value of fixationPeriod to specify the amount of time in milliseconds a subject has to look in a p
94. the position of the first line of tracked eye position and pupil diameter data This results data is delimited so that it may be readily imported into an Excel worksheet as shown below a oso 6 e Data O i84 fie Edit view Insert Format Tools Data RoboPDF Window Help 8X Arial 10 x BZ U amp E 192 fe A B c 7 1 DataStart 2 Line 12 3 Stimulus 4 Stimulus Still Picture vsg 5 Still Picture vsg B Duration 60 7 ImageFileName C Glasgow bmp 8 BackgroundColour 0 500 0 500 0 500 9 DataFormat 1D SeparatorChar 44 11 Tracked Horizontal Position Vertical Position Pupil Diameter DigitallO 12 1 13 494 0 13 1 13 545 2 61 5 6 0 14 1 13 6 2 615 5 61 n 15 1 13 561 2 638 561 0 46 1 13 605 2 68 56 0 17 1 13 638 2 647 5 61 0 18 1 13 613 2 644 5 6 0 19 1 13 602 2 638 5 59 0 20 1 13 67 2 678 5 62 0 2 1 13 667 2 643 5 62 0 22 1 13 623 2 63 5 61 0 23 1 13 624 2 608 5 6 0 24 1 13 658 2 648 5 61 0 25 1 13 616 2 633 5 61 0 26 1 13 637 2 605 5 61 27 1 13 653 2 607 5 62 0 28 1 13 695 2 613 5 62 0 29 1 13 758 2 592 5 62 n 30 1 13 708 2 671 5 61 0 31 1 13 686 2 664 5 62 0 1 T 2 61 5 62 0 v 5 gt MNSheet1 Sheet2 J 5666 zen lt Ready Fig 43 Results Data File 48 Video Eyetracker Toolbox videoEyetrace Custom Results Data File ved The custom data file option available from
95. their respective video pages whilst displaying a blank page to hide the drawing TargetSize crsDegreesToPixels 0 3 0 3 BlankPage 1 crsClearPage BlankPage GreyLevel FixationPage 2 crsClearPage FixationPage GreyLevel TargetPage 23 crsClearPage TargetPage GreyLevel crsSetDisplayPage BlankPage Draw a green fixation point on the fixation page crsSetDrawPage FixationPage crsSetPenl GreenLevel crsDrawOval 0 0 TargetSize Draw a red target on the target page crsSetDrawPage TargetPage crsSetPenl RedLevel crsDrawOval 0 0 TargetSize The next section of code defines the randomised timings and locations of the fixation and saccade target stimuli This is designed to prevent the test subject from anticipating the position or timing of the next stimulus This test will present a total of 10 horizontal fixation and saccade targets with displacements of 0 fixation 5 7 5 and 10 degrees Specify the timing for one trial We use a 1 3 second random interval to elicit reflexive rather than predictive saccades PageOrder FixationPage TargetPage FixationPage MinTime crsSecondsToFrames 0 5 1 0 5 MaxTime crsSecondsToFrames 1 5 3 1 5 Replicate the trials to provide data for all trials NumTrials 10 Pages repmat PageOrder 1 NumTrials MinTime repmat MinTime 1 NumTrials MaxTime repmat MaxTime 1 NumTrials Times roun
96. theoretical modelling and analysis The Toolbox s open interface design allows you to integrate eye tracking functionality with your existing laboratory software or write new eye tracking applications using a familiar environment Because the Toolbox has been designed in this way all aspects of its eye tracking functionality can be accessed and controlled from MATLAB or any windows programming tool that supports Microsoft COM 57 Video Eyetracker Toolbox Communication In particular MATLAB s relatively simple and concise scripting environment has gained a widespread following among the vision research community As a result MATLAB is now the default software environment for which Cambridge Research Systems provides support What follows are two worked examples which demonstrate the simplicity of using the Video Eyetracker Toolbox from within MATLAB The commands that make up the Video Eyetracker Toolbox interface for MATLAB are detailed in Appendix A If you want to use a different programming tool you will find more information about the COM interface in the CRS Support Portal www crsltd com support login Worked MATLAB Eye Tracking Examples The MATLAB scripts discussed in this section are distributed with the CRS Toolbox for MATLAB which should be downloaded from the Cambridge Research Systems Support Portal www crsltd com support login as described in Appendix A Example 1 Bitmap Stimulus with Dual VGA Example
97. tion To obtain accurate data regarding the diameter of your subject s pupil you need to calibrate the Toolbox so it knows the geometric scale of the video image it is being fed Once the correct scale factor has been set the Toolbox will be able to determine pupil diameter to within a few hundredths of a millimetre Note The scale factor does not have to be calibrated in order to track a subject s gaze position correctly To calibrate the scale factor for your subject follow the instructions below 1 Temporally remove the subject s head from the EyeLock headrest 2 Changing the focus point of the camera lens whilst positioning your subject will affect the scaling factor of the Toolbox 25 Video Eyetracker Toolbox Calibration 2 Click the Scale button on the Calibration window to enter pupil scale mode 3 Hold the Scale Calibration Stick in front of the camera lens so that it is in the same vertical plane that the subject s eye was located correctly focused and so one of the target dots appears in the centre of the Toolbox s camera window The target being presented to the Toolbox should be focused by altering the distance between the calibration stick and the mirror Do not adjust the camera s focusing ring The camera window should now contain an image similar to Fig 15 below Fig 15 Tracking the Pupil Scale Calibration Stick Whilst a target on the calibration stick is centred in the camera window
98. tly from the image data Subject calibration works as follows Subject Calibration subject calibration Save as Default Calibration Fig 13 Calibration Window with Default Settings The calibration window in Fig 13 can be displayed using one of the supplied demonstration scripts which are part of the CRS Toolbox for MATLAB These can be downloaded from our 1 Note these instructions assume the subject has been positioned correctly with their eye in focus and that the stimulus monitor has been calibrated for pixel size and viewing distance For more information consult the previous sections 24 Video Eyetracker Toolbox Calibration Support Portal www crsltd com support login see Appendix A on page 81 The calibration scripts are saved by default in the vetDemos folder You need to run either demoVETcalibrationVGA or demoVETcalibrationVSG depending on whether you have a Dual VGA or CRS Visual Stimulus Generator monitor setup This will create and display the Toolbox s Calibration window as shown in Fig 13 Note this window can also be accessed by clicking on the Calibrate icon Fig 14 within videoEyetrace If the Calibrate icon appears greyed out you need to make sure you have initialised your VET camera by choosing the appropriate type using the Select Video Source on the File menu see videoEyetrace chapter on page 35 for further details Fig 14 Calibrate Icon Pupil Scale Calibra
99. to match the conditions of the task that will be conducted by the subject after calibration To do this you should click on the Advanced button to display the Advanced Calibration Parameters window as shown in Fig 20 Using the options on this window it is possible to adjust the following characteristics These settings are listed in the order that they appear on the window Advanced Calibration Parameters Sy advanced calibration parameters Fixation Characteristics Duration 501 E ms Calibration Accuracy Level Calibration Points Vertical Points Horizontal Points Screen Scale Factor Target Stimulus Shape Circle Size 0 30 a Degrees Target Colour Background Colour Sy m Cancel Fig 20 Advanced Calibration Parameters Window e Fixation Duration This sets the time period data is collected over when making pupil and Purkinje measurements The larger this parameter the more frames will be used in the calculations and the longer the calibration procedure will take Calibration Accuracy level Allows you to adjust the tolerance the Toolbox uses when validating clusters of pupil and Purkinje measurements 30 Video Eyetracker Toolbox Calibration Vertical Points Calibration targets are arranged in a 2D array This setting determines how many vertical locations are used to perform the calibration Horizontal Points Same as above except this setting specifies the number of horizontal elements
100. ue stimulus presentation and eye movement analysis software GazeTracker is a powerful software package that consolidates stimulus presentation information synchronisation data analysis and visualisation See our website www crsltd com for further information Fig 2 GazeTracker Analysis of Dynamic Scene Content 11 Video Eyetracker Toolbox Introduction CRS Video Eyetracker Toolbox The Video Eyetracker Toolbox family of video eye trackers are designed to work with a standard PC running Windows XP without any custom processing hardware Apart from the obvious difference in the range of supported video frame rates from a user s perspective all systems share most of their functionality in common This document will therefore treat common functionality by default and only refer to specific systems where necessary 50 Hz Video Eyetracker Toolbox The 50 Hz Video Eyetracker Toolbox is a monocular eye tracker primarily intended for gaze tracking applications where the objective of the research is to determine where a test subject is looking and for how long The videoEyetrace picture in Fig 1 is an example of where the 50 Hz system has been used to track a horizontal smooth pursuit target while monitoring pupil diameter The 50 Hz systems are supplied with one of two different camera configurations called Standard Camera or Standard Camera l Standard Camera uses a progressive scan mode to acquire the video frames
101. using the vetCalibrateAdvanced function The vetCalibrate function will display targets on either a Secondary Windows VGA display or on a VSG depending on the stimulus device that has been selected using the vetSetStimulusDevice function If a user stimulus device has been selected then it is up to the user to draw fixation targets on their stimulus device under the control of user specifiable callbacks You can specify these callbacks using the following functions vetSetCallbacklnitialiseDisplay vetSetCallbackClearDisplay vetSetCallbackDrawTarget User device parameters also need to be set using the vetSetDeviceParameters function NOTE You must select the video source and the stimulus device before calling vetCalibrate See demoVETQuickstart for a worked example MATLAB Syntax vetCalibrate See also vetCalibrateAdvanced vetCalibrationStatus vetSelectVideoSource vetStartTracking vetLoadCalibrationFile vetSaveCalibrationFile 85 Video Eyetracker Toolbox Appendices vetCalibrateAdvanced Like vetCalibrate this function can be used to calibrate a subject vetCalibrateAdvanced opens the Calibration window but unlike vetCalibrate it passes a number of parameters to the window overwriting the Calibration window s default settings Thus this function makes it possible to perform a non standard calibration without using the Advanced Calibration Parameter dialog This function returns True 1 if the calibration has
102. vance beyond the first screen target during calibration will either be the result of the system failing to locate the Purkinje glints or the pupil or both The actual mode of failure will be shown in the live image window by the absence of one or more fitted ellipses see relevant section s below System fails to track the eye A sudden failure to track the eye will either be the result of the system failing to locate the Purkinje glints or the pupil or both The actual mode of failure will be shown in the live image window by the absence of one or more fitted ellipses see relevant section s below 139 Video Eyetracker Toolbox Appendices System fails to track the Pupil Failure of the system to locate the pupil could be due to one or more of the following e The pupil is no longer fully contained within the image or very close to the image edge Fig 75 Poorly Positioned Eye e The eye is no longer in focus due to head movement away from the infrared mirror Ensure that the subject maintains contact with the headrest at all times and refocus if necessary Fig 76 Out of Focus Eye 140 Video Eyetracker Toolbox Appendices e The eyelid is partially obscuring the eye Fig 77 Pupil Obscured by Eyelid e The eye is looking beyond its intended range of rotation If this occurs during calibration it may be necessary to reduce the screen scale factor in the calibration advanced settings window
103. vided they are either not as bright as the Purkinje reflections or do not lie close to the pupil You should now be able see an image of the subject s eye similar to Fig 12 Fig 12 Typical Video Image of Eye 9 Lastly before performing a calibration ask the subject to view the four corners of the stimulus screen Ensure that the two Purkinje glints remain compact and maintain constant separation over the full range of intended eye movements If not it will be necessary to reduce the range of intended eye rotations to ensure that both Purkinje glints always lie on the corneal surface This can be achieved by moving the stimulus screen further away from the eye or by changing the Screen 23 Video Eyetracker Toolbox Calibration Scale Factor in the Advanced Calibration Settings see next section Note if you move the stimulus display you must re measure the Viewing Distance and set the new value using the scheme appropriate to your graphics configuration see the Monitor Configuration section in this chapter for details Subject Calibration Calibration enables the Toolbox to learn the relationship between where a subject is looking and where their pupil and Purkinje glints appear in the video image This is achieved by asking your subject to fixate on a number of screen targets whose positions are known to the system Once calibration is complete the Toolbox will be able to calculate all subsequent viewpoints direc
104. w protocol Save protocol Calibrate subject Run stimulus presentation T BP I cti x A Boo v s Open protocol View options Load Trace zoom Stop stimulus and data file window stimulus in out presentation Fig 26 Toolbar 38 Video Eyetracker Toolbox videoEyetrace Data windows videoEyetrace outputs eye tracking results using two data windows The upper data window outputs the tracked eye positions in graphical form using the horizontal axis to represent time and the vertical axis to represent eye position During tracking the graph displays two traces showing how the horizontal and vertical components of eye position vary with time Untracked sections caused by blinking or looking outside the legitimate viewing range are denoted by a yellow trace superimposed on the horizontal time axis untracked 9 Horizontal screen position mm Vertical screen position mm Ii Fig 27 videoEyetrace Eye Tracking Graph The lower data window which shares the same horizontal time axis plots the calculated pupil width for each tracked frame plus a digital input trace if timing signals are used Pupil diameter mm IM DigitallO Fig 28 videoEyetrace Pupil Diameter and Digital O Graph 39 Video Eyetracker Toolbox videoEyetrace During tracking both graphical outputs will automatically scroll forwards and can be zoomed in or out either horizontally or vertically using the following toolbar icons
105. whereas Standard Camera l uses an interlaced mode Both will give equivalent performance but it will be necessary to select the appropriate type to match the supplied camera configuration before enabling the eye tracker Standard Cameras are supplied unlabelled whereas Standard Camera l versions are supplied with a label on the camera assembly High Speed Video Eyetracker Toolbox The High Speed version of the Video Eyetracker Toolbox is a monocular video eye tracker primarily intended for oculomotor research It provides a sampling frequency that is sufficient to capture the frequency spectrum of eye dynamics Fig 3 shows where videoEyetrace has been used with a 250 Hz sampling rate to measure eye movements during a saccade task while monitoring pupil diameter 12 Video Eyetracker Toolbox Introduction Re B E Bem z Selected Stimulus Saccade Task vso T Pipil Diameter 3 21 Calibrated Fig 3 High Speed Eyetracker with Saccade Stimulus Images Acquired with 50 Hz and High Speed Systems The major difference between the 50 Hz and High Speed versions of the Video Eyetracker Toolbox lies in the camera and illumination geometries required to operate at higher frame rates The 50 Hz system uses a standard CCD sensor whereas the High Speed version requires a CMOS sensor to operate at frequencies up to 250 Hz In addition the dual Purkinje glints generated by the infrared illu
106. ws you to limit the number of points returned by this function This is useful if the buffer is large and you can only deal with a limited number of points at any one time MATLAB Syntax data vetGetBufferedEyePositions 91 Video Eyetracker Toolbox Appendices data vetGetBufferedEyePositions removeFromBuffer data vetGetBufferedEyePositions maximumPoints data vetGetBufferedEyePositions maximumPoints removeFromBuffer Parameters maximumPoints Optional parameter A scalar integer indicating the maximum number of points to return to MATLAB removeFromBuffer Optional parameter A scalar logical controlling the removal of returned eye positions from the VET buffer Return Values data A MATLAB structure containing the VET eye position data Data timeStamps Time in milliseconds since tracking started Data tracked A Boolean value where 0 FALSE and 1 TRUE Data calibrated A Boolean value where 0 FALSE and 1 TRUE Data pupilDiameter Pupil diameter in millimetres Data mmPositions Direction of gaze projected onto stimulus display in mm X Y Data ficksPositions Eye rotation in Fick coordinates Longitude Latitude Data helmholtzPositions Eye rotation in Helmholtz coordinates Elevation Azimuth Data region Current region number Data fixation A Boolean value where 0 FALSE and 1 TRUE Data digitallO A value containing the combined digital IO values see page
107. ycle Pages Xlocations Ylocations Times T riggers Repeating Start eyetracking and stimulus cycling vetStartTracking crsSetCommand CRS CYCLEPAGEENABLE pause crsFramesToSeconds sum Times crsSetDisplayPage BlankPage vetStopTracking Tidy up by clearing away the camera screen vetDestroyCameraScreen vetDestroyMimicScreen Retrieve the recorded eye positions remove false DATA vetGetBufferedEyePositions remove Display the eye positions and triggers g 7 figure plot DATA ficksPositions 1 b longitude in blue hold on plot DATA ficksPositions 2 r latitude in red hold on plot DATA digitallO g triggers in green grid on Finally save the results to disk as a mat file tempfile C myResults mat vetSaveResults tempfile CRS ffMATfile 77 Video Eyetracker Toolbox Appendices d In this section Appendix A MATLAB Toolbox Commands 81 Appendix B Imaging and Illumination Geometry 122 Appendix C Template for Fixed Mount EyeLock 127 Appendix D System Specification 129 Appendix E Hot Mirror Frequency Response 131 Appendix F Fick and Helmholtz Coordinates 133 Appendix G Data Returned by the Toolbox 135 Appendix H Troubleshooting 137 Notes 147 79 Video Eyetracker Toolbox Appendices 80 Video Eyetracker Toolbox Appendices d Appendix A MATLAB Commantds The Video Eyetracker Toolbox can be controlled using MATLAB or any Windows prog
108. yetrace amount of time that the trace will persist before disappearing This is set by default to 5 seconds duration but can be altered accordingly The second dialog box controls whether the trace should persist with constant or fading intensity for the prescribed duration Video Eyetracker Options options Graph Configuration Mimic Window EIB Settings Trace Persistence 5 00 seconds Trace Style Constant v Trace Type Motion v Tace coo o C Fixation Colour Fig 36 Video Eyetracker Options Mimic Tab The third box controls the type of information to be plotted on the Mimic window This can be set to map out eye motion eye fixations or both Eye fixations are represented as circles where the centres represent eye position and the radii represent duration The last 2 boxes control the motion trace and fixation colours EIB Tab High Speed System only The EIB tab allows you to configure the High Speed Eyetracker Interface Box The first sub window allows you to disable the LEDs on the Eyetracker Interface Box front panel if for example their light output could possibly interfere with the experiment The 44 Video Eyetracker Toolbox videoEyetrace second sub window controls the signal output for the detected eye position This can be either 0 01 volts per mm 0 001 volts per mm or 0 1 volt per degree in either Fick or Helmholtz angular co ordinates Video Eyetracker Options
109. ze oe Specify the timing for one trial We use a 1 3 second random interval to elicit reflexive rather than predictive saccades PageOrder FixationPage TargetPage FixationPage MinTime crsSecondsToFrames 0 5 1 0 5 MaxTime crsSecondsToFrames 1 5 3 1 5 Replicate the trials to provide data for all trials NumTrials 10 Pages repmat PageOrder l NumTrials MinTime repmat MinTime 1 NumTrials MaxTime repmat MaxTime 1 NumTrials Times round MinTime rand MaxTime MinTime Specify target locations from 5 to 10 degrees We invert the x and y locations because we are moving the video window offset not the video memory Xlocations crsDegreesToPixels 5 0 7 5 10 0 Ylocations crsDegreesToPixels 0 0 0 0 0 0 Generate random locations for each entry Indices ceil rand size Pages numel Xlocations Xlocations Xlocations Indices Ylocations Ylocations Indices Make sure the fixation point always appears at 0 0 Xlocations Pages FixationPage 0 Ylocations Pages FixationPage 0 Specify triggers to identify when a target shift has occurred The VET captures digital output lines 6 7 and 8 Triggers zeros size Pages Triggers Pages FixationPage bin2dec 011100000 ole 76 Video Eyetracker Toolbox Communication o Set up RTS page cycling Repeating false crsSimplePageC

1 Video Eyetracker Toolbox

Contents

Download Pdf Manuals

Related Search

Related Contents