Video-based Realtime Eye Tracking Technique for Autostereoscopic Displays

An autostereoscopic display system can provide great enjoyment of stereo visualization without the uncomfortable and inconvenient drawbacks of wearing stereo glasses or HMD. In order to render the stereo video with respect to the user’s view point and to accurately project stereo video onto the eyes of the user, who is allowed to move around freely, the left and right eye positions of the user have to be obtained when the user is watching the autostereoscopic display. In this paper, we present a realtime eye tracking technique which can track the eye positions of the user in the video acquired with a camera. The user does not have to wear any sensor or mark and there is no restriction on the background. We employ a fast template matching technique to track the four motion parameters (X and Y translation, scaling, and rotation) of the user’s face in each image. The left and right eyes can then be located in the obtained face region. According to our implementation on a PC with Pentium III 500, the frame rate of the eye tracking process can achieve 30 Hz.