Eye-gaze Detection from Monocular Camera Image Based on Physiological Eyeball Models(International Workshop on Advanced Image Technology 2006)

概要

論文の詳細を見る
In the coming ubiquitous-computing society, an eye-gaze interface will be one of the key technologies as an input device. Most of the conventional eye-gaze tracking algorithms require specific light sources, equipments, devices, etc. In this paper, a simple eye-gaze detection algorithm is proposed using only a monocular video camera. First of all, in our system, the eyeball size of a user is estimated from the proposed 2-points calibration algorithm based on two kinds of physiological eyeball models. In the calibration process, the iris of the user is extracted from the captured face image by using the Hough transform. After the calibration process, the eye-gaze angle may be computed for each frame by detecting the position of the iris. By using the difference of the iris position, rotation degree of the eyeball can be calculated. Experimental results showed that the vertical error was 1.48[deg.] and the horizontal error was 1.65[deg.] on the average for 640×480[pixel] frames.
社団法人電子情報通信学会の論文
2006-01-02