Gazing-detection of human eyes based on SVM

Abstract

A method for gazing-detection of human eyes using Support Vector Machine (SVM) based on statistic learning theory (SLT) is proposed. According to the criteria of structural risk minimization of SVM, the errors between sample-data and model-data are minimized and the upper bound of predicting error of the model is also reduced. As a result, the generalization ability of the model is much improved. The simulation results show that, when limited training samples are used, the correct recognition rate of the tested samples can be as high as 100%, which is much better than some previous results obtained by other methods. The higher processing speed enables the system to distinguish gazing or not-gazing in real-time.

References

[1] HallinanP. W.. SPIE, 1991, 1570: 214-214

[2] A. Yuille, D. Cohen, and P. Hallinan.Proc. CVPR, 1989, 104.

[3] DengJyh-Yuan, LaiFeipei. Pattern Recognition, 1997, 30: 403-403

[4] XieX., SudhakarR., ZhuangH.. Pattern Recognition, 1994, 27: 791-791

[5] Athanasios Nikolaidis and loannis Pitas. Pattern Recognition, 2000, 33: 1783

[6] A. Gee and R. Cipolla.Proc. Mechatronics and Machine Vision in Practise, Toowoomba, Australia, (1994), 112.

[7] TalmiK., LiuJ.. Signal Processing: Image Communication, 1999, 14: 799-799

[8] YongWong, Jing-heYuan, Sheng-jiangChang, Yan-xinZhang. Journal of Optoelectronics · laser, 2001, 12: 1284-1284(in Chinese)

[9] YongWang, Jing-heYuan, Sheng-JiangChang, Yan-xinZhang. Optical Engineering, 2002, 41: 1840-1840

[10] D. A. Simon, M. Hebert, and Kanade.International Conference of Robotics and Automation Proceeding, May 94, San Diego, USA.