한국인터넷방송통신학회논문지 (The Journal of the Institute of Internet, Broadcasting and Communication)

  • 제13권2호
  • /
  • Pages.171-179
  • /
  • 2013
  • /
  • 2289-0238(pISSN)
  • /
  • 2289-0246(eISSN)
한국인터넷방송통신학회 (The Institute of Internet, Broadcasting and Communication)
권호 표지
DOI QR코드

DOI QR Code

강화학습을 이용한 눈동자 추적 시스템의 성능향상

Performance Improvement of Eye Tracking System using Reinforcement Learning

  • 신학철 (인하대학교 컴퓨터정보공학과) ;
  • 심연 (인하대학교 컴퓨터정보공학과) ;
  • 김사랑 (인하대학교 컴퓨터정보공학과) ;
  • 성원준 (인하대학교 컴퓨터정보공학과) ;
  • 민하즈 (인하대학교 컴퓨터정보공학과) ;
  • 홍요훈 (세창인스트루먼트) ;
  • 이필규 (인하대학교 컴퓨터정보공학과)
  • Shin, Hak-Chul (Dept. of Computer and Information Engineering, Inha University) ;
  • Shen, Yan (Dept. of Computer and Information Engineering, Inha University) ;
  • Khim, Sarang (Dept. of Computer and Information Engineering, Inha University) ;
  • Sung, WonJun (Dept. of Computer and Information Engineering, Inha University) ;
  • Ahmed, Minhaz Uddin (Dept. of Computer and Information Engineering, Inha University) ;
  • Hong, Yo-Hoon ;
  • Rhee, Phill-Kyu (Dept. of Computer and Information Engineering, Inha University)
  • 투고 : 2013.02.07
  • 심사 : 2013.04.12
  • 발행 : 2013.04.30
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

영상처리에서 인식에 관련된 기술들은 환경에 아주 많은 영향을 받게 되는데 이러한 인식률을 결정짓는 요소 중인 파라미터는 환경에 적절한 값을 얼마나 잘 선택하느냐에 따라서 인식률의 큰 차이를 보인다. 본 논문은 눈동자 추적 알고리즘이 사람이나 실험 환경의 변화에 따라 인식률이 저하되는 현상을 보완하기 위한 성능 향상 및 환경에 적응하는 시스템의 구현에 대한 방법이다. 최적의 파라미터를 얻기 위해 전 처리에 사용되는 이진화 알고리즘의 문턱값을 학습이 필요한 시기를 적절히 판단해 강화학습을 이용하여 다시 학습시켜 인식률을 향상시키는 방법을 사용했다. 실험데이터를 수집하기 위해 입력 장치는 가격이 저렴하고 일반적인 웹 카메라를 사용 하였으며 얼굴 영역에 해당하는 많은 양의 이미지를 수집하여 강화학습의 적응력을 실험하였다. 이미지의 그룹을 다양하게 변화시켜 실험한 결과 강화학습을 사용한 경우 그렇지 않은 경우에 비해 작게는 3% 많게는 14%가량의 성능이 향상됨을 확인하였다. 이렇게 성능이 향상된 눈동자 추적 시스템은 휴먼 컴퓨터 인터랙션 분야에 효과적으로 활용될 수 있을 것이다.

Recognition and image processing technology depends on illumination variation. One of the most important factors is the parameters of algorithms. When it comes to select these values, the system has different types of recognition accuracy. In this paper, we propose performance improvement of the eye tracking system that depends on some environments such as, people, location, and illumination. Optimized threshold parameter was decided by using reinforcement learning. When the system accuracy goes down, reinforcement learning used to train the value of parameters. According to the experimental results, the performance of eye tracking system can be improved from 3% to 14% by using reinforcement learning. The improved eye tracking system can be effectively used for human-computer interaction.

키워드

참고문헌

  1. Allen, M., Fritzsche, P., "Reinforcement learning with adaptive Kanerva coding for Xpilot game AI", Evolutionary Computation (CEC), 2011 IEEE Congress on, pp. 1521-1528, 2011.
  2. Anton Leuski, "Relevance and Reinforcement in Interactive Browsing", Center for Intelligent Information Retrieval, Department of Computer Science University of Massachusetts, USA.
  3. Young-Woo Seo and Byoung-Tak Zhang, "A reinforcement learning agent for personalized information filtering", Proceedings of the 2000 international conference on Intelligent user interfaces, pp. 248-251, 2000.
  4. Ueno, A., Tateyama, T., Takase, M., Minamitani, H., "Bluntness of saccadic eye movement depending on vigilance states: examination by model simulation", Engineering in Medicine and Biology Society, 1997. Proceedings of the 19th Annual International Conference of the IEEE, pp.1482-1484, 1997.
  5. Iwamoto, K., Tanie, K., "High resolution, wide view angle head mounted display using eye movement tracking: system structure and evaluation experiments", Robot and Human Communication, 1995. RO-MAN'95 TOKYO, Proceedings., 4th IEEE International Workshop on, pp.289-294, 1995.
  6. Kierkels, J.J.M., Riani, J., Bergmans, J.W.M., van Boxtel, G.J.M., "Using an Eye Tracker for Accurate Eye Movement Artifact Correction", Biomedical Engineering, pp.1256-1267, 2007
  7. Hiley, J.B., Redekopp, A.H., Fazel-Rezai, R., "A Low Cost Human Computer Interface based on Eye Tracking", Engineering in Medicine and Biology Society, 2006. EMBS '06. 28th Annual International Conference of the IEEE, pp.3226-3229, 2006.
  8. Model, D., Eizenman, M., "An Adaptive Algorithm for Eye-Gaze-Tracking-Device Calibration", Electrical and Computer Engineering (CCECE), 2011 24th Canadian Conference on, pp.1268-1271, 2011.
  9. Model, D., Guestrin, E.D., Eizenman, M., "An automatic calibration procedure for remote eye-gaze tracking systems", Engineering in Medicine and Biology Society, 2009. EMBC 2009. Annual International Conference of the IEEE, pp.4751-4754, 2009.
  10. Lili Pan, Wen-Sheng Chu, Jason M. Saragih, "Fernando De la Torre, Fast and Robust Circular Object Detection With Probabilistic Pairwise Voting", IEEE SIGNAL PROCESSING LETTERS, VOL. 18, NO. 11, NOVEMBER 2011.
  11. Kuo-Liang Chung, Yong-HuaiHuang, Jyun- PinWanga, Ting-Chin Chang, Hong-YuanMarkLiao, "Fast randomized algorithm for center-detection", Pattern Recognition, 2010.
  12. Liang, D.B.B., Lim Kok Houi, "Non-intrusive eye gaze direction tracking using color segmentation and Hough Transform", Communications and Information Technologies, 2007. ISCIT '07. International Symposium on, pp. 602-607, 2007.

피인용 문헌

  1. The Study on Dynamic Position Control base on Neural Networks, Image Processing and CAN Communication vol.17, pp.11, 2013, https://doi.org/10.6109/jkiice.2013.17.11.2499
  2. A Study on Adaptive Skin Extraction using a Gradient Map and Saturation Features vol.15, pp.7, 2014, https://doi.org/10.5762/KAIS.2014.15.7.4508