基于规则与卡尔曼滤波的人眼跟踪

Eye Tracking Based on Rules and the Kalman Filtering

本文提出一种基于YCbCr空间、规则和卡尔曼滤波的人眼跟踪方法。该方法首先利用肤色在YCbCr空间的特性建立肤色模型分割出肤色,然后通过设定的规则得到人眼区域,最后利用卡尔曼滤波的方法实时跟踪人眼。实验表明,该方法能达到实时效果且跟踪正确率较高。

An eye tracking method based on the YCbCr color space, rules and the Kalman filtering is proposed in this paper. First, all skin color regions can be got from a skin model in the YCbCr color space, then the eye regions can be found according to some rules. Finally, the eye position is tracked by the Kalman filtering. Experimental results show that the method can track the eye in a real-time way and has very high correctness rate.