SlideShare una empresa de Scribd logo
1 de 58
Descargar para leer sin conexión
A Real-Time Automatic Eye Tracking
              System for Ophthalmology
                                          Mr. Wattanit Hotrakool
                                         Mr. Prarinya Siritanawan
                                Supervised by Dr.Toshiaki Kondo
                  Sirindhorn International Institute of Technology




1
Outline
       Project Background
       Project Objective
       A eye-tracking technique using Traditional Template
        Matching
       A eye-tracking technique using Gradient Orientation
        Pattern Matching
       A eye-tracking technique using Time-varying GOPM
       Conclusion
       Question and Answer



    2
Project Background
       The conventional eye-surgery cameras are manual-
        controlled and they reduce the efficiency of surgery.
       In order to reduce the burden of oculist, the automated
        camera control is required. The image processing is used
        to locate and track the eye’s centroid.




    3
Project Background
       Many real-time eye tracking techniques used intensity
        data as an input; they are very sensitive to changing
        lighting condition and result as miss-matching.
       This project proposes new template matching based
        technique which robust to changing lighting condition by
        using Time-varying Gradient         Orientation  Pattern
        Matching.




    4
Project Objectives
       To implement matching-based techniques in real-time.
       To verify the robustness to changing lighting condition of
        gradient orientation pattern matching.
       To develop new eye-tracking technique; a time-varying
        gradient orientation pattern matching.




    5
Eye-tracking method using template
matching technique
       Template matching is the intensity-based technique for
        measuring the similarity between template and
        corresponding block of image.




              Template

                                                  Match




                               Sample frame

    6
Simulation Specification

   Simulation System using the following hardware and
    software specifications:

                          Software Specification
              Operating System              Windows/Linux
            Programming Language                   C/C++
               Primary library               OpenCV 2.0
                         Hardware Specification
                 Processor                 Intel Core2 Duo
               Processor Speed                1.66 GHz
                  Memory                           4GB


    7
Video sequences used for simulation
 Video sequences used in this simulation can be
  categorized into 3 categories:
1. Test video in normal lighting condition
2. Test video in changing lighting condition
3. Actual surgery video from real camera




    8
Template Initialization
       Template initialization is required before the eye-tracking
        method.
       The pupil is specified as a template in order to track the
        eye.

       The example of video and corresponding template for
        each categories is shown in next page




    9
Test video in normal lighting condition




               Video sequence




                  Template
10
Test video in changing lighting condition




               Video sequence




                  Template
11
Actual surgery video from real camera




               Video sequence




                 Template
12
A eye-tracking method using
     Traditional template matching
13
A eye-tracking method using Traditional
template matching
    There are many traditional techniques of template
     matching such as sum-of-absolute difference (SAD), sum-
     of-squared difference (SSD), or cross-correlation (CC)
     technique.

    In this step, we implement the sum-of-squared difference
     technique (SSD) to be eye-tracking method.




    14
Sum-of-squared Difference technique
    SSD is the template matching method done by finding the
     lowest difference value between input and template. The
     differences are squared in order to remove the sign.




            where I1 is the intensity of input image and
            I2 is the intensity of template
            N is the size of the template



    15
Procedure
1.    Convert template to gray scale image.
2.    Convert input frame to gray scale image.
3.    For every pixel, compute the SSD between input and
      template.
4.    Find the minimum difference pixel, which is the best
      matching location.
5.    For every frame, repeat step 2-4.




 16
Procedure




17
Result
        Input              Average computation time      Precision error
                                     (ms)                      (%)
   Normal light video                78.35                     1.33
(resolution: 320x240 px)
  Changing light video               78.63                     40.47
(resolution: 320x240 px)
  Actual surgery video              103.33                      0
(resolution: 384x288 px)


        Average computation time mostly depends on video
                          resolution

     However, this method currently can process at 10 -12.5 frame/sec



18
Result
        Input             Average computation time          Precision error
                                    (ms)                          (%)
   Normal light video                 78.35                       1.33
 (resolution: 320x240)
  Changing light video                78.63                      40.47
 (resolution: 320x240)
 Actual surgery video                 103.33                       0
 (resolution: 384x288)


         SSD technique can work very well in normal light video.
       However, this technique give high error in changing light video
         because it uses intensity data which are sensitive to light.

      Therefore SSD cannot work in changing light condition


19
Result
             Properties of eye-tracking using SSD technique
       Obstacle robustness                           Yes
          Blur robustness                            Yes
     Light condition robustness                      No
         Scaling robustness                          No
     Average computation time                  About 50-350 ms




20
A eye-tracking method using Gradient
     Orientation Pattern Matching
               Presented at ICESIT2010, Chiang Mai, Thailand
21
A eye-tracking technique using Gradient
Orientation Pattern Matching
    In order to develop a method that can provide the
     robustness to light condition, the new template matching
     technique is used.

    The gradient orientation pattern matching (GOPM) is a
     new template matching technique proposed by Dr.
     Toshiaki Kondo.




    22
Gradient Orientation Pattern Matching
    GOPM is a template matching method which use the
     normalized gradient of the image in place of intensity
     data. Thus, it only consider about the shape of the
     pattern but not light.
    Gradient vector is the first derivative of intensity. The
     gradient in x and y direction are defined as:




         where I is a intensity of an image

    23
Gradient Orientation Pattern Matching
    The gradient in x and y will then be normalized. This
     step provides the robustness to light condition. The
     normalized gradient in x and y direction are defined
     as:




                   where
            And   is a small constant used to prevent zero-division.

    24
Gradient Orientation Pattern Matching
    The normalized gradient in x and y direction of input
     frame and template will be match by using SSD




         where N1 is the normalized gradient of input image and
         N2 is the normalized gradient of template


    25
Procedure
 For every frame, we can divide the procedure into 2 main
  steps;
1. Gradient orientation information (GOI) extraction
2. Gradient orientation pattern matching (GOPM)




    26
Gradient Orientation Information (GOI)
Extraction

                          Extract the gradient Images
                          (Template and sample frame) in
                          x and y direction.




27
Gradient Orientation Pattern Matching




                    Apply template matching in x and y
                    direction. Then add the result of x and
                    y direction




28
Result
          Input                 SSD technique            GOPM technique
                              Average     Precision     Average     Precision
                            computation   error (%)   computation   error (%)
                             time (ms)                 time (ms)
Normal lighting condition      78.35        1.33         62.48          0
(resolution: 320x240px)
Change lighting condition      78.63       40.47         63.09        12.87
(resolution: 320x240px)
 Actual surgery camera        103.33         0           77.7           0
(resolution: 384x288px)

       Average computation time of GOPM slightly inprove from
                               SSD

       Even though the method is more complex, but the computation time
      is decrease due to the variable type and internal structure of OpenCV
           However, this method still can process at only <15 frame/sec
 29
Result
           Input                SSD technique             GOPM technique
                              Average      Precision     Average      Precision
                            computation    error (%)   computation    error (%)
                             time (ms)                  time (ms)
Normal lighting condition      78.35         1.33         62.48          0
(resolution: 320x240px)
Change lighting condition      78.63        40.47         63.09        12.87
(resolution: 320x240px)
 Actual surgery camera        103.33          0            77.7          0
(resolution: 384x288px)

      In changing light condition, GOPM error is dramatically decrease due to
                               the normalized process.

      Therefore GOPM can provide the robustness to changing light
                             condition

 30
Result

       Properties of eye-tracking using SSD and GOPM technique
                                         SSD               GOPM
       Obstacle robustness                Yes                Yes
          Blur robustness                 Yes                Yes
     Light condition robustness           No                 Yes
         Scaling robustness               No                 No
     Average computation time       About 50-350 ms    About 50-350 ms




31
A eye-tracking method using a
             Time-varying GOPM
        Accepted by ECTI-CON 2010, Chiang Mai, Thailand

32
Time-varying GOPM
    Even though GOPM provides robustness to changing light
     condition, however the static template will not
     guaranteed that it yields the good result for all condition.
     There are many uncontrolled factors such as skin and
     noise.
    Time-varying GOPM uses the dynamic template which
     update itself automatically in place of static template. It
     reduce the difference of template environment in various
     period of time.




    33
Template Update Algorithm
    Step 1 : Perform GOPM, get best matching coordinate



                                                current Template



    BEST MATCH




                           Sample frame


    34
Template Update Algorithm
    Step 2 : Crop region with the same size of old template
     for creating new template




                                                  new Template



                            Sample frame


    35
Correct-matching criterion
    1st Criterion Equation




            where Nxn+1 and Nyn+1 are the normalized gradient of the newly created
            template, Nxn and Nyn are the normalized gradient of the current
            template, i and j are the size of the template.
              is a threshold value defined as




        1st Criterion is used to check the correctness of the updated
         template and prevent the jumping coordinate.


    36
Correct-matching criterion
    2nd Criterion Equation


            where Xn and Yn are the location of current matching result,
            Xc and Yc are the location of the last known correct result.
            T is a threshold value define as




        2nd Criterion is used to double check the jumping real
         coordinate.


    37
1st Criterion
    STEP 1 : Find the difference b/w gradient component of
     old template and new template in X and Y direction


x
                        -                  =
y
                        -                  =
         Old Template       New Template          Template diff.



    38
1st Criterion
    STEP 2 : Combine the difference of x and y


                        +                        =
           Diff x                  Diff y                    Total Diff


    STEP 3 : Sum all elements and then thresholding
        If summation is less than the threshold function, update
         template.
        If summation is more than the threshold function, discard the
         new template.

    39
31

2nd Criterion

    Using the fact that It is impossible that the eye would
     change its position suddenly in next frame.




                                             (400,100)
               (300,500)




                  Frame N                  Frame N+1


40
2nd Criterion
    STEP 1 : Find the best matching of the frame N.




         (x1,y1)




                             Frame N


    41
2nd Criterion
    STEP 2 : If location of N passed the criteria, the location
     is used as a latest known correct position C.




                              Frame N


    42
2nd Criterion
    STEP 3 : Find best matching of the frame N+1.



                                                     (x2,y2)




                            Frame N+1


    43
2nd Criterion
    STEP 4 : Find Euclidean distance between position C and
     position in frame N+1.




                            Frame N+1


    44
2nd Criterion
     STEP 5.1 : If distance more than threshold function,
      discard the current location.




Threshold fcn



                              Frame N+1


     45
2nd Criterion
     STEP 5.2 : If distance less than threshold function, mask
      the location as new corrected position C.




Threshold fcn



                              Frame N+1


     46
Correct-matching criteria with Template
Update




47
Procedure




48
Downsampling
    In here, we resizes input video sequence and template to
     50 percent of the height and width. Hence the
     downsampled image is reduced to ¼ of the original size.
    Thus, computation time is 4 times faster.
    No effect to the matching result since both video and
     template are downsampled with the same ratio




    49
Result
         Input                 GOPM technique          Time-varying GOPM
                              Average     Precision     Average     Precision
                            computation   error (%)   computation   error (%)
                             time (ms)                 time (ms)
Normal lighting condition      62.48         0           13.81         0
(resolution: 320x240 px)
Change lighting condition      63.09       12.87         12.92         0
(resolution: 320x240 px)
 Actual surgery camera         77.7          0          17.12          0
(resolution: 384x288 px)


       Average computation time is decreased by downsampling

      Currently this method can process at > 50 frame/sec which is enough
          for most of video capture device that run at 25 frame/secs.
 50
Result
         Input                 GOPM technique          Time-varying GOPM
                              Average     Precision     Average     Precision
                            computation   error (%)   computation   error (%)
                             time (ms)                 time (ms)
Normal lighting condition      62.48         0           13.81         0
(resolution: 320x240 px)
Change lighting condition      63.09       12.87         12.92         0
(resolution: 320x240 px)
 Actual surgery camera         77.7          0          17.12          0
(resolution: 384x288 px)

    In all cases, time-varying GOPM provide the better result than normal
   GOPM. Especially for the case of changing light condition, which error is
                                decrease to 0%
    Therefore time-varying GOPM can provide the robustness to
  changing light condition with more precision than normal GOPM
 51
Result

 Properties of eye-tracking using SSD, GOPM , and time-varying GOPM
                                technique
                                         SSD          GOPM         Time-varying
         Obstacle robustness             Yes            Yes            Yes
           Blur robustness               Yes            Yes            Yes
     Lighting condition robustness       No             Yes            Yes
          Scaling robustness             No             No             No
      Average computation time       About 50-350   About 50-350   About 10-90
                                         ms             ms             ms




52
Drawback of time-varying GOPM
    In rarely case, when the 2nd criteria drop the frame
     repeatedly, it causes the template slightly shifts from the
     eye’s centroid.



                                Fail update



    However for real implementation prospective, if not be
     too much, can be tolerated by the surgeon



    53
Conclusion




54
Conclusion
    This work verify that the speed of template matching
     technique with downsampling is able to implement in
     real-time. (speed > 50 frame/secs).
    In the changing light condition, the result clearly shows
     that GOPM is more robust than SSD.
    A time-varying GOPM reduce the difference of
     template environment in various time and provides the
     higher precision of tracking than normal GOPM.




    55
Future Work
    Optimize the utilization of the threshold function in
     corrected-matching criterion.
    Due to difference in camera specification such as
     resolution or sensitivity, it required other advance
     method to supervised the threshold function such as
     machine learning or neuron network.




    56
Acknowledgement
    Assist. Prof. Dr. Toshiaki Kondo
    Assoc.Prof. Dr. Waree Kongprawechnon
    Dr. Itthisek Nilkhamhang
    All faculty members and our beloved friends




    57
Question and Answer




58

Más contenido relacionado

La actualidad más candente

Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...
Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...
Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...IJECEIAES
 
Survey on Image Integration of Misaligned Images
Survey on Image Integration of Misaligned ImagesSurvey on Image Integration of Misaligned Images
Survey on Image Integration of Misaligned ImagesIRJET Journal
 
IRJET-Retina Image Decomposition using Variational Mode Decomposition
IRJET-Retina Image Decomposition using Variational Mode DecompositionIRJET-Retina Image Decomposition using Variational Mode Decomposition
IRJET-Retina Image Decomposition using Variational Mode DecompositionIRJET Journal
 
DNW_apriori_error est_report
DNW_apriori_error est_reportDNW_apriori_error est_report
DNW_apriori_error est_reportAnkur Kislaya
 
International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)IJERD Editor
 
IRJET- Retinal Health Diagnosis using Image Processing
IRJET- Retinal Health Diagnosis using Image ProcessingIRJET- Retinal Health Diagnosis using Image Processing
IRJET- Retinal Health Diagnosis using Image ProcessingIRJET Journal
 
Tomosurgery Presentation
Tomosurgery PresentationTomosurgery Presentation
Tomosurgery Presentationngowdar
 
Automatic dynamic depth focusing for ndt
Automatic dynamic depth focusing for ndtAutomatic dynamic depth focusing for ndt
Automatic dynamic depth focusing for ndtJose Miguel Moreno
 
Image Compression based on DCT and BPSO for MRI and Standard Images
Image Compression based on DCT and BPSO for MRI and Standard ImagesImage Compression based on DCT and BPSO for MRI and Standard Images
Image Compression based on DCT and BPSO for MRI and Standard ImagesIJERA Editor
 
Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display
Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display  Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display
Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display ijcga
 
IRJET - A Review on Gradient Histograms for Texture Enhancement and Objec...
IRJET -  	  A Review on Gradient Histograms for Texture Enhancement and Objec...IRJET -  	  A Review on Gradient Histograms for Texture Enhancement and Objec...
IRJET - A Review on Gradient Histograms for Texture Enhancement and Objec...IRJET Journal
 
An Efficient Thresholding Neural Network Technique for High Noise Densities E...
An Efficient Thresholding Neural Network Technique for High Noise Densities E...An Efficient Thresholding Neural Network Technique for High Noise Densities E...
An Efficient Thresholding Neural Network Technique for High Noise Densities E...CSCJournals
 
Implementation of Lower Leg Bone Fracture Detection from X Ray Images
Implementation of Lower Leg Bone Fracture Detection from X Ray ImagesImplementation of Lower Leg Bone Fracture Detection from X Ray Images
Implementation of Lower Leg Bone Fracture Detection from X Ray Imagesijtsrd
 
Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...
Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...
Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...IJMERJOURNAL
 
The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)theijes
 
Purkinje imaging for crystalline lens density measurement
Purkinje imaging for crystalline lens density measurementPurkinje imaging for crystalline lens density measurement
Purkinje imaging for crystalline lens density measurementPetteriTeikariPhD
 

La actualidad más candente (20)

Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...
Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...
Neuroendoscopy Adapter Module Development for Better Brain Tumor Image Visual...
 
Survey on Image Integration of Misaligned Images
Survey on Image Integration of Misaligned ImagesSurvey on Image Integration of Misaligned Images
Survey on Image Integration of Misaligned Images
 
IRJET-Retina Image Decomposition using Variational Mode Decomposition
IRJET-Retina Image Decomposition using Variational Mode DecompositionIRJET-Retina Image Decomposition using Variational Mode Decomposition
IRJET-Retina Image Decomposition using Variational Mode Decomposition
 
DNW_apriori_error est_report
DNW_apriori_error est_reportDNW_apriori_error est_report
DNW_apriori_error est_report
 
Sirona Orthophos X 3 G
Sirona  Orthophos X 3 GSirona  Orthophos X 3 G
Sirona Orthophos X 3 G
 
Kc3118711875
Kc3118711875Kc3118711875
Kc3118711875
 
International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)
 
IRJET- Retinal Health Diagnosis using Image Processing
IRJET- Retinal Health Diagnosis using Image ProcessingIRJET- Retinal Health Diagnosis using Image Processing
IRJET- Retinal Health Diagnosis using Image Processing
 
Tomosurgery Presentation
Tomosurgery PresentationTomosurgery Presentation
Tomosurgery Presentation
 
Automatic dynamic depth focusing for ndt
Automatic dynamic depth focusing for ndtAutomatic dynamic depth focusing for ndt
Automatic dynamic depth focusing for ndt
 
Image Compression based on DCT and BPSO for MRI and Standard Images
Image Compression based on DCT and BPSO for MRI and Standard ImagesImage Compression based on DCT and BPSO for MRI and Standard Images
Image Compression based on DCT and BPSO for MRI and Standard Images
 
Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display
Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display  Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display
Augmented Reality in Volumetric Medical Imaging Using Stereoscopic 3D Display
 
IRJET - A Review on Gradient Histograms for Texture Enhancement and Objec...
IRJET -  	  A Review on Gradient Histograms for Texture Enhancement and Objec...IRJET -  	  A Review on Gradient Histograms for Texture Enhancement and Objec...
IRJET - A Review on Gradient Histograms for Texture Enhancement and Objec...
 
An Efficient Thresholding Neural Network Technique for High Noise Densities E...
An Efficient Thresholding Neural Network Technique for High Noise Densities E...An Efficient Thresholding Neural Network Technique for High Noise Densities E...
An Efficient Thresholding Neural Network Technique for High Noise Densities E...
 
An Image-Based Bone fracture Detection Using AForge Library
An Image-Based Bone fracture Detection Using AForge LibraryAn Image-Based Bone fracture Detection Using AForge Library
An Image-Based Bone fracture Detection Using AForge Library
 
Implementation of Lower Leg Bone Fracture Detection from X Ray Images
Implementation of Lower Leg Bone Fracture Detection from X Ray ImagesImplementation of Lower Leg Bone Fracture Detection from X Ray Images
Implementation of Lower Leg Bone Fracture Detection from X Ray Images
 
Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...
Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...
Rician Noise Reduction with SVM and Iterative Bilateral Filter in Different T...
 
The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)
 
Purkinje imaging for crystalline lens density measurement
Purkinje imaging for crystalline lens density measurementPurkinje imaging for crystalline lens density measurement
Purkinje imaging for crystalline lens density measurement
 
False colouring
False colouringFalse colouring
False colouring
 

Similar a A real time automatic eye tracking system for ophthalmology

THE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERA
THE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERATHE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERA
THE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERANational Cheng Kung University
 
NEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODING
NEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODINGNEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODING
NEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODINGcscpconf
 
2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘
2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘
2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘CHENHuiMei
 
Realtime pothole detection system using improved CNN Models
Realtime pothole detection system using improved CNN ModelsRealtime pothole detection system using improved CNN Models
Realtime pothole detection system using improved CNN Modelsnithinsai2992
 
Machine Learning approaches at video compression
Machine Learning approaches at video compression Machine Learning approaches at video compression
Machine Learning approaches at video compression Roberto Iacoviello
 
Iec 62676 5 standardized spezification of image quality
Iec 62676 5 standardized spezification of image qualityIec 62676 5 standardized spezification of image quality
Iec 62676 5 standardized spezification of image qualityHenry Rutjes
 
A Novel Approach for Tracking with Implicit Video Shot Detection
A Novel Approach for Tracking with Implicit Video Shot DetectionA Novel Approach for Tracking with Implicit Video Shot Detection
A Novel Approach for Tracking with Implicit Video Shot DetectionIOSR Journals
 
International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)IJERD Editor
 
Disparity Estimation by a Real Time Approximation Algorithm
Disparity Estimation by a Real Time Approximation AlgorithmDisparity Estimation by a Real Time Approximation Algorithm
Disparity Estimation by a Real Time Approximation AlgorithmCSCJournals
 
Recognition and tracking moving objects using moving camera in complex scenes
Recognition and tracking moving objects using moving camera in complex scenesRecognition and tracking moving objects using moving camera in complex scenes
Recognition and tracking moving objects using moving camera in complex scenesIJCSEA Journal
 
Google | Infinite Nature Zero Whitepaper
Google | Infinite Nature Zero WhitepaperGoogle | Infinite Nature Zero Whitepaper
Google | Infinite Nature Zero WhitepaperAlejandro Franceschi
 
Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...
Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...
Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...CSCJournals
 
Multimodel Operation for Visually1.docx
Multimodel Operation for Visually1.docxMultimodel Operation for Visually1.docx
Multimodel Operation for Visually1.docxAROCKIAJAYAIECW
 
Poster_ARVO_ADCIS_AGN-V6a
Poster_ARVO_ADCIS_AGN-V6aPoster_ARVO_ADCIS_AGN-V6a
Poster_ARVO_ADCIS_AGN-V6aJoe Buonomo
 
Color Detection & Segmentation based Invisible Cloak
Color Detection & Segmentation based Invisible CloakColor Detection & Segmentation based Invisible Cloak
Color Detection & Segmentation based Invisible CloakAviral Chaurasia
 
Image super resolution using Generative Adversarial Network.
Image super resolution using Generative Adversarial Network.Image super resolution using Generative Adversarial Network.
Image super resolution using Generative Adversarial Network.IRJET Journal
 
Paper 58 disparity-of_stereo_images_by_self_adaptive_algorithm
Paper 58 disparity-of_stereo_images_by_self_adaptive_algorithmPaper 58 disparity-of_stereo_images_by_self_adaptive_algorithm
Paper 58 disparity-of_stereo_images_by_self_adaptive_algorithmMDABDULMANNANMONDAL
 
Design and Analysis of Quantization Based Low Bit Rate Encoding System
Design and Analysis of Quantization Based Low Bit Rate Encoding SystemDesign and Analysis of Quantization Based Low Bit Rate Encoding System
Design and Analysis of Quantization Based Low Bit Rate Encoding Systemijtsrd
 

Similar a A real time automatic eye tracking system for ophthalmology (20)

THE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERA
THE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERATHE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERA
THE 3D MODELLING USING FRAME CAMERAS AND PANORAMIC CAMERA
 
NEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODING
NEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODINGNEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODING
NEW IMPROVED 2D SVD BASED ALGORITHM FOR VIDEO CODING
 
2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘
2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘
2018AOI論壇_深度學習於表面瑕疪檢測_元智大學蔡篤銘
 
Realtime pothole detection system using improved CNN Models
Realtime pothole detection system using improved CNN ModelsRealtime pothole detection system using improved CNN Models
Realtime pothole detection system using improved CNN Models
 
Machine Learning approaches at video compression
Machine Learning approaches at video compression Machine Learning approaches at video compression
Machine Learning approaches at video compression
 
Iec 62676 5 standardized spezification of image quality
Iec 62676 5 standardized spezification of image qualityIec 62676 5 standardized spezification of image quality
Iec 62676 5 standardized spezification of image quality
 
Main1
Main1Main1
Main1
 
A Novel Approach for Tracking with Implicit Video Shot Detection
A Novel Approach for Tracking with Implicit Video Shot DetectionA Novel Approach for Tracking with Implicit Video Shot Detection
A Novel Approach for Tracking with Implicit Video Shot Detection
 
International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)
 
Disparity Estimation by a Real Time Approximation Algorithm
Disparity Estimation by a Real Time Approximation AlgorithmDisparity Estimation by a Real Time Approximation Algorithm
Disparity Estimation by a Real Time Approximation Algorithm
 
Recognition and tracking moving objects using moving camera in complex scenes
Recognition and tracking moving objects using moving camera in complex scenesRecognition and tracking moving objects using moving camera in complex scenes
Recognition and tracking moving objects using moving camera in complex scenes
 
Google | Infinite Nature Zero Whitepaper
Google | Infinite Nature Zero WhitepaperGoogle | Infinite Nature Zero Whitepaper
Google | Infinite Nature Zero Whitepaper
 
Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...
Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...
Tracking Chessboard Corners Using Projective Transformation for Augmented Rea...
 
Multimodel Operation for Visually1.docx
Multimodel Operation for Visually1.docxMultimodel Operation for Visually1.docx
Multimodel Operation for Visually1.docx
 
Poster_ARVO_ADCIS_AGN-V6a
Poster_ARVO_ADCIS_AGN-V6aPoster_ARVO_ADCIS_AGN-V6a
Poster_ARVO_ADCIS_AGN-V6a
 
Color Detection & Segmentation based Invisible Cloak
Color Detection & Segmentation based Invisible CloakColor Detection & Segmentation based Invisible Cloak
Color Detection & Segmentation based Invisible Cloak
 
NMSL_2017summer
NMSL_2017summerNMSL_2017summer
NMSL_2017summer
 
Image super resolution using Generative Adversarial Network.
Image super resolution using Generative Adversarial Network.Image super resolution using Generative Adversarial Network.
Image super resolution using Generative Adversarial Network.
 
Paper 58 disparity-of_stereo_images_by_self_adaptive_algorithm
Paper 58 disparity-of_stereo_images_by_self_adaptive_algorithmPaper 58 disparity-of_stereo_images_by_self_adaptive_algorithm
Paper 58 disparity-of_stereo_images_by_self_adaptive_algorithm
 
Design and Analysis of Quantization Based Low Bit Rate Encoding System
Design and Analysis of Quantization Based Low Bit Rate Encoding SystemDesign and Analysis of Quantization Based Low Bit Rate Encoding System
Design and Analysis of Quantization Based Low Bit Rate Encoding System
 

Último

Potential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsPotential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsRavi Sanghani
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality AssuranceInflectra
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.Curtis Poe
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersNicole Novielli
 
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...panagenda
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch TuesdayIvanti
 
Testing tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesTesting tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesKari Kakkonen
 
The State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxThe State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxLoriGlavin3
 
Manual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditManual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditSkynet Technologies
 
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024BookNet Canada
 
Rise of the Machines: Known As Drones...
Rise of the Machines: Known As Drones...Rise of the Machines: Known As Drones...
Rise of the Machines: Known As Drones...Rick Flair
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demoHarshalMandlekar2
 
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better StrongerModern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better Strongerpanagenda
 
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptxMerck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptxLoriGlavin3
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024Lonnie McRorey
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxLoriGlavin3
 
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...Scott Andery
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxLoriGlavin3
 
The Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsThe Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsPixlogix Infotech
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI AgeCprime
 

Último (20)

Potential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsPotential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and Insights
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software Developers
 
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch Tuesday
 
Testing tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesTesting tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examples
 
The State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxThe State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptx
 
Manual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditManual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance Audit
 
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
 
Rise of the Machines: Known As Drones...
Rise of the Machines: Known As Drones...Rise of the Machines: Known As Drones...
Rise of the Machines: Known As Drones...
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demo
 
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better StrongerModern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
 
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptxMerck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
 
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
 
The Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsThe Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and Cons
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI Age
 

A real time automatic eye tracking system for ophthalmology

  • 1. A Real-Time Automatic Eye Tracking System for Ophthalmology Mr. Wattanit Hotrakool Mr. Prarinya Siritanawan Supervised by Dr.Toshiaki Kondo Sirindhorn International Institute of Technology 1
  • 2. Outline  Project Background  Project Objective  A eye-tracking technique using Traditional Template Matching  A eye-tracking technique using Gradient Orientation Pattern Matching  A eye-tracking technique using Time-varying GOPM  Conclusion  Question and Answer 2
  • 3. Project Background  The conventional eye-surgery cameras are manual- controlled and they reduce the efficiency of surgery.  In order to reduce the burden of oculist, the automated camera control is required. The image processing is used to locate and track the eye’s centroid. 3
  • 4. Project Background  Many real-time eye tracking techniques used intensity data as an input; they are very sensitive to changing lighting condition and result as miss-matching.  This project proposes new template matching based technique which robust to changing lighting condition by using Time-varying Gradient Orientation Pattern Matching. 4
  • 5. Project Objectives  To implement matching-based techniques in real-time.  To verify the robustness to changing lighting condition of gradient orientation pattern matching.  To develop new eye-tracking technique; a time-varying gradient orientation pattern matching. 5
  • 6. Eye-tracking method using template matching technique  Template matching is the intensity-based technique for measuring the similarity between template and corresponding block of image. Template Match Sample frame 6
  • 7. Simulation Specification  Simulation System using the following hardware and software specifications: Software Specification Operating System Windows/Linux Programming Language C/C++ Primary library OpenCV 2.0 Hardware Specification Processor Intel Core2 Duo Processor Speed 1.66 GHz Memory 4GB 7
  • 8. Video sequences used for simulation  Video sequences used in this simulation can be categorized into 3 categories: 1. Test video in normal lighting condition 2. Test video in changing lighting condition 3. Actual surgery video from real camera 8
  • 9. Template Initialization  Template initialization is required before the eye-tracking method.  The pupil is specified as a template in order to track the eye.  The example of video and corresponding template for each categories is shown in next page 9
  • 10. Test video in normal lighting condition Video sequence Template 10
  • 11. Test video in changing lighting condition Video sequence Template 11
  • 12. Actual surgery video from real camera Video sequence Template 12
  • 13. A eye-tracking method using Traditional template matching 13
  • 14. A eye-tracking method using Traditional template matching  There are many traditional techniques of template matching such as sum-of-absolute difference (SAD), sum- of-squared difference (SSD), or cross-correlation (CC) technique.  In this step, we implement the sum-of-squared difference technique (SSD) to be eye-tracking method. 14
  • 15. Sum-of-squared Difference technique  SSD is the template matching method done by finding the lowest difference value between input and template. The differences are squared in order to remove the sign. where I1 is the intensity of input image and I2 is the intensity of template N is the size of the template 15
  • 16. Procedure 1. Convert template to gray scale image. 2. Convert input frame to gray scale image. 3. For every pixel, compute the SSD between input and template. 4. Find the minimum difference pixel, which is the best matching location. 5. For every frame, repeat step 2-4. 16
  • 18. Result Input Average computation time Precision error (ms) (%) Normal light video 78.35 1.33 (resolution: 320x240 px) Changing light video 78.63 40.47 (resolution: 320x240 px) Actual surgery video 103.33 0 (resolution: 384x288 px) Average computation time mostly depends on video resolution However, this method currently can process at 10 -12.5 frame/sec 18
  • 19. Result Input Average computation time Precision error (ms) (%) Normal light video 78.35 1.33 (resolution: 320x240) Changing light video 78.63 40.47 (resolution: 320x240) Actual surgery video 103.33 0 (resolution: 384x288) SSD technique can work very well in normal light video. However, this technique give high error in changing light video because it uses intensity data which are sensitive to light. Therefore SSD cannot work in changing light condition 19
  • 20. Result Properties of eye-tracking using SSD technique Obstacle robustness Yes Blur robustness Yes Light condition robustness No Scaling robustness No Average computation time About 50-350 ms 20
  • 21. A eye-tracking method using Gradient Orientation Pattern Matching  Presented at ICESIT2010, Chiang Mai, Thailand 21
  • 22. A eye-tracking technique using Gradient Orientation Pattern Matching  In order to develop a method that can provide the robustness to light condition, the new template matching technique is used.  The gradient orientation pattern matching (GOPM) is a new template matching technique proposed by Dr. Toshiaki Kondo. 22
  • 23. Gradient Orientation Pattern Matching  GOPM is a template matching method which use the normalized gradient of the image in place of intensity data. Thus, it only consider about the shape of the pattern but not light.  Gradient vector is the first derivative of intensity. The gradient in x and y direction are defined as: where I is a intensity of an image 23
  • 24. Gradient Orientation Pattern Matching  The gradient in x and y will then be normalized. This step provides the robustness to light condition. The normalized gradient in x and y direction are defined as: where And is a small constant used to prevent zero-division. 24
  • 25. Gradient Orientation Pattern Matching  The normalized gradient in x and y direction of input frame and template will be match by using SSD where N1 is the normalized gradient of input image and N2 is the normalized gradient of template 25
  • 26. Procedure  For every frame, we can divide the procedure into 2 main steps; 1. Gradient orientation information (GOI) extraction 2. Gradient orientation pattern matching (GOPM) 26
  • 27. Gradient Orientation Information (GOI) Extraction Extract the gradient Images (Template and sample frame) in x and y direction. 27
  • 28. Gradient Orientation Pattern Matching Apply template matching in x and y direction. Then add the result of x and y direction 28
  • 29. Result Input SSD technique GOPM technique Average Precision Average Precision computation error (%) computation error (%) time (ms) time (ms) Normal lighting condition 78.35 1.33 62.48 0 (resolution: 320x240px) Change lighting condition 78.63 40.47 63.09 12.87 (resolution: 320x240px) Actual surgery camera 103.33 0 77.7 0 (resolution: 384x288px) Average computation time of GOPM slightly inprove from SSD Even though the method is more complex, but the computation time is decrease due to the variable type and internal structure of OpenCV However, this method still can process at only <15 frame/sec 29
  • 30. Result Input SSD technique GOPM technique Average Precision Average Precision computation error (%) computation error (%) time (ms) time (ms) Normal lighting condition 78.35 1.33 62.48 0 (resolution: 320x240px) Change lighting condition 78.63 40.47 63.09 12.87 (resolution: 320x240px) Actual surgery camera 103.33 0 77.7 0 (resolution: 384x288px) In changing light condition, GOPM error is dramatically decrease due to the normalized process. Therefore GOPM can provide the robustness to changing light condition 30
  • 31. Result Properties of eye-tracking using SSD and GOPM technique SSD GOPM Obstacle robustness Yes Yes Blur robustness Yes Yes Light condition robustness No Yes Scaling robustness No No Average computation time About 50-350 ms About 50-350 ms 31
  • 32. A eye-tracking method using a Time-varying GOPM  Accepted by ECTI-CON 2010, Chiang Mai, Thailand 32
  • 33. Time-varying GOPM  Even though GOPM provides robustness to changing light condition, however the static template will not guaranteed that it yields the good result for all condition. There are many uncontrolled factors such as skin and noise.  Time-varying GOPM uses the dynamic template which update itself automatically in place of static template. It reduce the difference of template environment in various period of time. 33
  • 34. Template Update Algorithm  Step 1 : Perform GOPM, get best matching coordinate current Template BEST MATCH Sample frame 34
  • 35. Template Update Algorithm  Step 2 : Crop region with the same size of old template for creating new template new Template Sample frame 35
  • 36. Correct-matching criterion  1st Criterion Equation where Nxn+1 and Nyn+1 are the normalized gradient of the newly created template, Nxn and Nyn are the normalized gradient of the current template, i and j are the size of the template. is a threshold value defined as  1st Criterion is used to check the correctness of the updated template and prevent the jumping coordinate. 36
  • 37. Correct-matching criterion  2nd Criterion Equation where Xn and Yn are the location of current matching result, Xc and Yc are the location of the last known correct result. T is a threshold value define as  2nd Criterion is used to double check the jumping real coordinate. 37
  • 38. 1st Criterion  STEP 1 : Find the difference b/w gradient component of old template and new template in X and Y direction x - = y - = Old Template New Template Template diff. 38
  • 39. 1st Criterion  STEP 2 : Combine the difference of x and y + = Diff x Diff y Total Diff  STEP 3 : Sum all elements and then thresholding  If summation is less than the threshold function, update template.  If summation is more than the threshold function, discard the new template. 39
  • 40. 31 2nd Criterion  Using the fact that It is impossible that the eye would change its position suddenly in next frame. (400,100) (300,500) Frame N Frame N+1 40
  • 41. 2nd Criterion  STEP 1 : Find the best matching of the frame N. (x1,y1) Frame N 41
  • 42. 2nd Criterion  STEP 2 : If location of N passed the criteria, the location is used as a latest known correct position C. Frame N 42
  • 43. 2nd Criterion  STEP 3 : Find best matching of the frame N+1. (x2,y2) Frame N+1 43
  • 44. 2nd Criterion  STEP 4 : Find Euclidean distance between position C and position in frame N+1. Frame N+1 44
  • 45. 2nd Criterion  STEP 5.1 : If distance more than threshold function, discard the current location. Threshold fcn Frame N+1 45
  • 46. 2nd Criterion  STEP 5.2 : If distance less than threshold function, mask the location as new corrected position C. Threshold fcn Frame N+1 46
  • 47. Correct-matching criteria with Template Update 47
  • 49. Downsampling  In here, we resizes input video sequence and template to 50 percent of the height and width. Hence the downsampled image is reduced to ¼ of the original size.  Thus, computation time is 4 times faster.  No effect to the matching result since both video and template are downsampled with the same ratio 49
  • 50. Result Input GOPM technique Time-varying GOPM Average Precision Average Precision computation error (%) computation error (%) time (ms) time (ms) Normal lighting condition 62.48 0 13.81 0 (resolution: 320x240 px) Change lighting condition 63.09 12.87 12.92 0 (resolution: 320x240 px) Actual surgery camera 77.7 0 17.12 0 (resolution: 384x288 px) Average computation time is decreased by downsampling Currently this method can process at > 50 frame/sec which is enough for most of video capture device that run at 25 frame/secs. 50
  • 51. Result Input GOPM technique Time-varying GOPM Average Precision Average Precision computation error (%) computation error (%) time (ms) time (ms) Normal lighting condition 62.48 0 13.81 0 (resolution: 320x240 px) Change lighting condition 63.09 12.87 12.92 0 (resolution: 320x240 px) Actual surgery camera 77.7 0 17.12 0 (resolution: 384x288 px) In all cases, time-varying GOPM provide the better result than normal GOPM. Especially for the case of changing light condition, which error is decrease to 0% Therefore time-varying GOPM can provide the robustness to changing light condition with more precision than normal GOPM 51
  • 52. Result Properties of eye-tracking using SSD, GOPM , and time-varying GOPM technique SSD GOPM Time-varying Obstacle robustness Yes Yes Yes Blur robustness Yes Yes Yes Lighting condition robustness No Yes Yes Scaling robustness No No No Average computation time About 50-350 About 50-350 About 10-90 ms ms ms 52
  • 53. Drawback of time-varying GOPM  In rarely case, when the 2nd criteria drop the frame repeatedly, it causes the template slightly shifts from the eye’s centroid. Fail update  However for real implementation prospective, if not be too much, can be tolerated by the surgeon 53
  • 55. Conclusion  This work verify that the speed of template matching technique with downsampling is able to implement in real-time. (speed > 50 frame/secs).  In the changing light condition, the result clearly shows that GOPM is more robust than SSD.  A time-varying GOPM reduce the difference of template environment in various time and provides the higher precision of tracking than normal GOPM. 55
  • 56. Future Work  Optimize the utilization of the threshold function in corrected-matching criterion.  Due to difference in camera specification such as resolution or sensitivity, it required other advance method to supervised the threshold function such as machine learning or neuron network. 56
  • 57. Acknowledgement  Assist. Prof. Dr. Toshiaki Kondo  Assoc.Prof. Dr. Waree Kongprawechnon  Dr. Itthisek Nilkhamhang  All faculty members and our beloved friends 57