Eye Tracking Based on Grey Prediction
Jianxiong Tang, Jianxin Zhang · 2009
Eye tracking is the focus problem in the researching domain of human-machine interaction and computer vision in recent years. The method of using a single eye location and detection algorithm has poor real-time performance. So a new eye tracking method is proposed in this paper. This method combines the location and detection algorithm with the grey prediction for eye tracking. The GM(1,1) model is used to predict the position of moving eye in the next frame, and then this position is taken as the reference point for the searching region of eye. The efficiency of eye tracking is improved in this way. Experimental results for image sequence of eye maneuvering show that the grey prediction model GM(1,1) can track eye region robustly and correctly because it doesnpsilat assume the moving law in advance, and weaken the effect of random disturbances by results of accumulated generating operation to explore the law of eye motion.