驾驶困倦预警系统中眼部状态识别的研究

为判断驾驶员是否处于疲劳驾驶状态,通过对人脸正面图像中的眼部状态的研究,提出了一种判断算法.首先基于灰度投影曲线和人脸几何特点提取驾驶员的眼部区域,然后利用事先提取的驾驶员睁眼时的瞳孔区域模板在该区域进行匹配,通过匹配初步判断出驾驶员是否属于疲劳驾驶.实验结果表明,该算法简单高效,在满足一定外部条件下,误判率较低.

Study on Recognition of the Eye States for Early Alarming Systems of Driver Fatigue

Driver fatigue is one of the most important reasons of the traffic accidents,and the variance of eyes is very evident.An eye states recognition approach is proposed in this paper.Firstly,the eye area is identified from the face image based on the curves of gray-level projection and geometrical characteristics of human face;then the obtained eyes area is compared with the preparative eye template to judge whether the eye is closed or not.Experiments show that the proposed method is simple and effective,and the error probability is rather small in certain environmental conditions.