A fast eye-movement tracking using cross-division algorithm

This study shows that a precise eye tracking and pupillometer system, neither interfering visual field nor influenced by head movement, can be implemented by analyzing first Paikinje image of visual field which has a few reference points. If two or more reference points are sited around the center of visual field, then eye camera captures the overlapped image of reference points and pupil. Four high intensity infra-red light-emitting diodes (IR LEDs) are used as reference points in this study not to disturb the subject's attention. These reference points appear as four highlighting points in the captured pupil image. The actual gazing point can be calculated by finding the distance between the center of these four reference points and pupil center and angle. Head movement causes some shift of location of pupil and reference points in the same direction and with the same amount hence does not change the relation of them. This technique can afford to make a system that has fast sampling time and high resolution without extra servo-mechanism to chase subject's head movement.