Docstoc

EYE TRACKING AND DETECTION BY USING FUZZY TEMPLATE MATCHING AND PARAMETER BASED JUDGMENT

Document Sample
EYE TRACKING AND DETECTION BY USING FUZZY TEMPLATE MATCHING AND PARAMETER BASED JUDGMENT Powered By Docstoc
					  International Journal of              Engineering and Technology (IJCET), ISSN 0976-
 INTERNATIONALComputer VolumeOF COMPUTER ENGINEERING
                              JOURNAL 4, Issue 1, January- February (2013), © IAEME
  6367(Print), ISSN 0976 – 6375(Online)
                             & TECHNOLOGY (IJCET)
ISSN 0976 – 6367(Print)
ISSN 0976 – 6375(Online)
Volume 4, Issue 1, January- February (2013), pp. 80-88
                                                                              IJCET
© IAEME: www.iaeme.com/ijcet.asp
Journal Impact Factor (2012): 3.9580 (Calculated by GISI)                  ©IAEME
www.jifactor.com




   EYE TRACKING AND DETECTION BY USING FUZZY TEMPLATE
        MATCHING AND PARAMETER BASED JUDGMENT
                        1
                            TARUN DHAR DIWAN, 2UPASANA SINHA
                                    ASSISTANT PROFESSOR
                                    DEPT. OF ENGINEERIN
                      1
                        Dr.C.V.RAMAN UNIVERSITY, BILASPUR (INDIA)
                  2
                    J K INSTITUTE OF ENGINEERING, BILASPUR (INDIA)
                        1
                          taruncsit@gmail.com, 2upasana.sihna@gmail.com


  ABSTRACT

          The eyes are tracked and correlation scores between the actual eye and the
  corresponding “closed-eye” template are used to detect blinks. In which a fuzzy template is
  constructed based on the piecewise boundary. A judgment of eye or non eye is made
  according to the similarity between input image and eye template. Eye blinking is one of the
  prominent areas to solve many real world problems. The work that has been carried out for
  eye tracking only is not suitable for eye blink detection. Stored template for a particular depth
  is chosen. Once the template is chosen and the system is in operation, the subject will be
  restricted to be at the specified distance. Another disadvantage of the system is that changing
  camera Positions require the whole system to be retrained the process of blink detection
  consists of two phases. These are eye tracking followed by detection of blink. The work that
  has been carried out for eye tracking only is not suitable for eye blink detection. Therefore
  some approaches had been proposed for eye tracking along with eyes blink detection. This
  paper implements one of the approaches given by Michael et al [1]. Further more the result of
  template creation accuracy and total blink detection to count number of eye blinks in an
  image sequence. Online template is completely independent of any past templates that may
  have been created during the run of the system.

  Keywords - template, frames, Interface, testing, automatically, involuntary.

  1. INTRODUCTION

         There has been a growing interest in the field of facial expression recognition
  especially in the last two decades. The primary contribution of this research is automatically

                                                 80
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME

initializing the eye blink detection in an image sequence for real time eye tracking applications. The
never ending saga of traffic accidents all over the world are due to deterioration of driver’s vigilance
level. Drivers with a depleting vigilance level suffer from a marked decline in their perception;
recognition and vehicle control abilities & therefore pose a serious danger to their own lives and the
lives of the other people. For this reason, developing systems that actively monitors the driver’s level
of vigilance and alerting the driver of any insecure driving condition is essential for accident
prevention [2]. Many efforts have been reported in the literature for developing an active safety
system for reducing the number of automobiles accidents due to reduced vigilance. Though advance
safety features are provided such as advances in vehicle design, including the provision of seat belts
and airbags and improvements in crashworthiness have led to considerable casualty reductions in
recent years [3].However, future increases in road traffic will. Make it difficult to meet future casualty
reduction targets unless more advanced accident avoidance technologies can be introduced.

2. RELATED WORK

         Lots of works have been carried out to detect face and extract features from it. Main facial
muscles that correspond to facial changes are eyebrow raiser, eyebrow frowning, lip suck and eye
blink [4]. Whenever we talk about eye blinking, tracking of eyes become built in need. Lots of
approaches have been developed to track eyes. Kanade et al. [5] proposed a method to locate eyes in
static images which was improved & re implemented several times. Kanade et al. [6] have shown that
the approaches used for eye tracking cause error in case of eye blinking if it is incorporated into an
image sequence. Therefore, Farhan et al. [7] proposed eye tracker along with blink detection
algorithm. Here they first detect the face by using motion based head segmentation. For tracking the
eye, they used the inner corner of eyes as invariant property because this property is invariant towards
the lighting changes. Variance map of frames of image sequence and statistical operation on
connected components were used to detect the eye blink. It also used normalized correlation
coefficient to detect eye blink. This coefficient is insensitive to lighting condition so it gives better
result.

3. METHODS

         The algorithm used by the system for counting the eye blinking in the video taken by USB
camera is initialized automatically, dependent only upon the inevitability of the involuntary blinking
of the user. Motion analysis techniques are used in this stage, followed by online creation of a
template of the open eye to be used for the subsequent tracking and template matching that is carried
out at each frame. [2, 8, 9]A flow chart depicting the main stages of the system is shown in




                Figure 1. Flow chart of the Approach of eye blinks detection.

                                                   81
 International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
 6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME


 The first step in counting the blinking of the user is to locate the eyes. For this, the difference
 of two subsequent frames is taken and then thresholding is done. The resulting image shows
 the regions of movement that occurred between the two frames. Next to remove the noise of
 background movement, an Opening morphological operation is performed by using diamond
 shape structuring element.
 The reflection of light on the surface of glasses makes the diarized eyes small pieces of
 disconnected areas, so we used eyebrows as the Primary feature to locate sunglasses [10]. The
 eyebrow is one of visible and stable features of the face, so it can be used as a secondary
 feature in sunglasses detection. The accurate position of eyebrows will determine the accuracy
 of detecting sunglasses. To locate eyebrow as one separate region, the main difficulty is that
 when the driver turns aside, the area representing eye or eyebrow may be connected with dark
 parts surrounding the head, which makes it impossible to search such an area. So, we need to
 split the desired connected region apart from its surrounding [11, 12].




                       Figure2. Gray image and resulting binary image

   We try to locate the region that is likely to contain primary features (such as eyes, eyebrows)
based on Then connected components in the resultant image is found and labeled. For
discarding the other movement except eye blinking, filtering of unlikely eye pair is done which
is based on the computation of six parameters for each component pair: the width and height of
each of the two components and the horizontal and vertical distance between the centroids of
the two components [13, 14]. Thus after this process, eye pair is received if present otherwise
steps are continued for other subsequent frames.




                  Figure 3. Shows the images of the output of above process.

After locating the eye pair, a template of 55x55 size of one of the eye is created. For detecting
the eye blink, normalized correlation function is used in each frame of the video which gives
the value between 0 and 1. Then the maximum value of the correlation coefficient is taken
from each frame. If its value is greater than 0.94 then eye is considered open otherwise close.
So the counting of close eye frames is done to count the number of times the eye is blinked.
                                                82
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME

4. TEMPLATE CREATION

         If the previous stage results in a pair of components that passes the set of filters, then it is a
good indication that the user’s eyes have been successfully located. At this point, the location of the
larger of the two components is chosen for creation of the template. Since the size of the template that
is to be created is directly proportional to the size of the chosen component, the larger one is chosen
for the purpose of having more brightness information, which will result in more accurate tracking
and correlation scores[15].




                                    Figure 4. Open eye templates

Since the system will be tracking the user’s open eye, it would be a mistake to create the template at
the instant that the eye was located, since the user was blinking at this moment. Thus, once the eye is
believed to be located, a timer is triggered. After a small number of frames elapse, which is judged to
be the approximate time needed for the user’s eye to become open again after an involuntary blink,
the template of the user’s open eye is created [1, 3, 16]. Therefore, during initialization, the user is
assumed to be blinking at a normal rate of one involuntary blink every few moments. Again, no
offline templates are necessary and the creation of this online template is completely independent of
any past templates that may have been created during the run of the system.

5. Experiments

Table 1 Results of Template Generation Accuracy, Automatic Blink Detection, Manually Blink
Detection, Missed Blink Detection.


  Automatic Blink Detection             Manually Blinks Detection            Missed Blink detection

                167                                   151                                31
                184                                   167                                24
                181                                   156                                25
                139                                   123                                 9
                176                                   170                                24
               Total                                 Total                              total
                847                                   767                               113


                                                    83
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME

6. RESULT AND DISSCUSATION

       Table 2 Results of Template Generation Accuracy, Accuracy of Blink Detection.



 Template Generation Accuracy         Find out Error     Accuracy of Template matching


               75 %                       10.59%                      89.41 %
               80 %                       10.18%                      89.82 %
               80 %                       16.02%                      83.98 %

               75 %                      13.008%                      86.99 %
               85 %                       3.52%                       96.47 %

               Total                       Total                       Total

                395                        53.318%                   446.67%

               Avg.                        Avg.                        Avg.
               79%                        10.66%                      89.33%

For experiment, total 100 videos are used in different lightning condition using inbuilt USB
camera of Samsung RV 509. The size of each frame is 480x640. The result of template
creation accuracy and total blink detection is tabulated in Table 1 for each video.847
automatic template creation is achieved and 79% accuracy and 89.33% accuracy of template
detection is achieved in counting of eye blink for 100 videos. The result may be tested for
more number of videos.




                               Figure 5. Template Generation



                                             84
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME

Template Generation:We have ploted a pia chart for template generation in which you can
see he percentage of tem1plate generation is ploted by different colour.As shown video 1to
20 is navy blue representing 19% of template generation, the maroom colour represents video
from 21 to 40 consisting of 20% present of template generation then comes video 41 to 61
which is dark in colour an consists of template generation then from video 61 to 80 is
reresented by violet colour consisting of 19% template generation and lastly from video 81 to
100 consists of 22% of template generation and is represent by sky blu colour




                        Figure 6. Accuracy of Template matching

for more convience we have also ploted of graph shown template generation accuracy in this
we have divided it into 6 bars in which 5 bars shown the video from 1to 100 and 6th bar show
the overall template generation from 1 to 20 the accuracy 75% from 21 to 40 the accracy is
80% from video 41 to 60 it is 80% from video 61 to 80 it is 75% from the video81 to 100 it is
85% and the over all template generation accuracy is 79%.




                                  Figure 7. Template Detection

Template Detection : We have plotted a pia chart for template detection in which you can see
he percentage of template detection is ploted by different colour.As shown video 1to 20 is
navy blue representing 89.41% of template detection, the maroom colour represents video
from 21 to 40 consisting of 89.82% present of template detection then comes video 41 to 61
consisting of 83.98% which is dark in colour an consists of template detection then from

                                             85
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME

video 61 to 80 is reresented by violet colour consisting of 86.99% template detection and
lastly from video 81 to 100 consists of 96.47% of template detection and is represent by sky
blu colour.




                        Figure 8. Accuracy of Template Detection


for more convince we have also plotted of graph shown template detection accuracy in this
we have divided it into 6 bars in which 5 bars shown the video from 1to 100 and 6th bar show
the overall template detection from 1 to 20 the accuracy 89.41% from 21 to 40 the accracy is
89.82% from video 41 to 60 it is 83.98% from video 61 to 80 it is 86.99% from the video81
to 100 it is 96.47% and the over all template detection accuracy is 89%.

7. CONCLUSION

        After studying and analyzing results of above technique following points is
concluded:
1. A good accuracy is achieved in different illumination conditions. Testing must be done on
large database of videos.
2.over all template generation accuracy is 79% and template detection accuracy is 89%.
 3. The initialization technique is efficient and gives good results. The system responds
slowly and requires more work for real time implementation.

8. APPLICATION AREA

   Automobiles.
   Security Guard Cabins.
   Operators at nuclear power plants where continuous monitoring is necessary.
   Pilots of airplane.
   Military application where high intensity monitoring of soldier is needed.
   Medical sectors for Eye related problems.
   Personal identification system



                                            86
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME

REFERENCES
[1]      Tarun Dhar Diwan "Real Time Eye Template generation system in an image
sequence",CIIT International Journal Of Digital Image Processing, June2012, ISSN: 0974-9586,
DOI: DIP062012013.
[2]      Tarun Dhar Diwan“Automatic Eye Blink Tracking & Detection in an Image Sequence"
InternationalJournal of Computer Science and Information Technology, Vol- 2 , pages 2348-2349
2011, ISSN 0975 – 9646.
[3]      Tarun Dhar Diwan"Improve Frame Generacy Accuracy With USB Cameras" International
Journal of Electronics and Computer science Engineering, Volume1, 2012, pages 1427-1432, ISSN
2277-1956.
[4]       Tarun Dhar Diwan "Eye Tracking and Detection by Using Template Generation and
 Parameter Based Judgment"CiiT International Journal Of Digital Image Processing, August 2012
 ,ISSN: 0974-9586, DOI: DIP.082012006
[5] Tarun Dhar Diwan "Local Binary Pattern Occuence Map Method for High Parallel Image
Processing" International Conference on Advances in Computing and Communication Aprl 8-10,
2011, pages 538-540, ISBN:978-81-920874-0-5, IEEE,NIT Hamirpur, Himachal Pradesh, India
[6] Tarun Dhar Diwan "personal identification system ", CiiT - International Journal of Data Mining
Knowledge Engineering, June2012, and ISSN: 0974-9578, DOI: DMKE062012004.
[7] Y. Tian, T. Kanade & J. Cohn 1999. "Multi-State Based Facial Feature Tracking and
Detection.Robotics Institute", Carnegie Mellon University, Technical Report CMU-RI-TR-99-18.
[8]. X. Xie, R. Sudhakar & H. Zhuang. On Improving Eye Feature Extraction Using Deformable
Templates. Pattern Recognition 27,pages 791-799,1994.
[9]. A. L. Yuille, D. S. Cohen & P. W. Hallinan. Feature Extraction from Faces Using Deformable
Templates. Proceedings Computer Vision and Pattern Recognition, pages 104-109, 2008.
[10] X. Wei, Z. Zhu, L. Yin, and Q. Ji. A real-time face tracking and animation system. Proceedings
of the CVPR Workshop on Face Processing in Video (FPIV2004),Washington, D.C., June 28 2004.
[11] M. Betke,W. Mullally, and J. Magee. Active detection of eye scleras in real time. Proceedings of
the IEEE CVPR Workshop on Human Modeling, Analysis and Synthesis (HMAS 2000), Hilton Head
Island, SC, June2000.
[12] T.N. Bhaskar, F.T. Keat, S. Ranganath, and Y.V.Venkatesh. Blink detection and eye tracking for
eye localization. Proceedings of the Conference on Convergent Technologies for Asia-Pacific Region
(TENCON2003), pages 821–824, Bangalore, Inda, October15-17 2003.
[13] S. Crampton and M. Betke. Counting fingers in real time:A webcam-based human-computer
interface with game applications. Proceedings of the Conference on Universal Access in Human-
Computer Interaction (affiliated with HCI International 2003), pages1357–1361, Crete, Greece, June
2003.
[14]. J. Deng and F. Lai." Region based template deformation and masking for eye feature extraction
and description". Pattern Recognition, pages 403–419, 1997.
[15] D.O. Gorodnichy. On importance of nose for face tracking. Proceedings of the IEEE
International Conference on Automatic Face and Gesture Recognition (FG 2002), pages 188–196,
Washington, D.C., May 20-21 2002.
[16] “Digital Image Processing Using MATLAB” Rafael C. Gonzalez, Richard E.Woods,Steven
L.Eddins, Mc Graw Hill, Second Edition, 2010.
[17] Mr.J.Rajarajan, Dr.G.Kalivarathan, “Influence Of Local Segmentation In The Context Of Digital
Image Processing – A Feasibility Study”, International journal of Computer Engineering &
Technology (IJCET), Volume 3, Issue 3, 2012, pp. 340 - 347, Published by IAEME.




                                                 87
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 1, January- February (2013), © IAEME




AUTHOR BIOGRAPHIES




Tarun Dhar Diwan received his Master of Engineering (Computer Technology and
Application) degree from Chhattisgarh swami Vivekananda technical university –Bhilai,
India, and Master of Philosophy (Gold Medal list) from Dr. C.V. Raman University. He is
currently HOD & Mtech Coordinate at the Dr.C.V.Raman University-bilaspur, India. His
Current research work artificial intelligent, Image Processing and Software Engineering.




                                           88

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:12
posted:2/2/2013
language:
pages:9