Blink Detection Using 3D Cross Model
Yifei Xu, Yibin Jiang, Yaoru Sun · 2012
In this paper, we present a new method based on a novel 3D cross model for detecting human eye blinks. Firstly, an initial frontal face is given to constructe the 3D cross model, which represents the head. by using the method of optical flow, the 3D cross model can track the head motion in real-time without the effects of self-occlusion and head large motion. According to the horizontal information of 3D cross model, the position of eyes can be located dynamically and accurately in each frame. the number of pixels whose gray values less than a certain threshold is calculated when the eye is open and close respectively. by comparing the different nums, the eye's status can be detected. Experimental results show that the method can efficiently achieved an overall accuracy of 99.24% in real-time.