A Robust Eye Detection and Tracking Technique Using Gabor Filters
Yen‐Wei Chen, Kenji Kubo · 2007
Eye detection is an important issue in the area of face recognition and vision based human-machine interface and so on. In this paper, we propose a robust eye (pupil) detection and tracking technique using Gabor filters. We first use Gabor filter and skin color information to detect the face and then an eye candidate region is automatically determined based on the geometric structure of human face. We then four Gabor filters with different directions (0, pi/4, pi/2, 3pi/4) to the eye candidate region. Since the pupil of eye do not have any directions, the pupil or the eye can be easily detected by combining the four responses of Gabor filters with a logical product. The proposed method is able to robustly detect and track eyes, and is insensitive to distance, pose and glass. In addition, the proposed method is also able to simultaneously detect and track multiple people.