ABSTRACT
Firstly we come to two-dimensional visual tracking. This tracking method uses a mapping function which has been calibrated to estimate the gazing direction. The import of the mapping function comes from a series of eye-movement characteristics for each eyes. And the export of the mapping function is the gazing direction of the sight point. The eye-movement characteristics changes with eye-sight direction. In order to get the mapping function, we need to calibrate each user on line. But two-dimensional visual tracking has two disadvantages. One is that the precision of the calibration decrease with the user’s head motion, so we should stabilize the user’s head. Another is that the calibration should test for many times.
