基于Adaboost算法的人脸疲劳检测

眼睛状态是人体疲劳最主要和最明显的特征.本文采用肤色和Adaboost方法相结合来进行人脸检测,并在此基础上结合人脸结构的边缘特征及Adaboost方法对眼睛进行精确定位,运用自适应二值化和数学形态学的方法对检测出的图像进行处理提取眼睛状态特征,结合PERCLOS规则及点头率来进行疲劳状态的判定,实验表明,该方法鲁棒性强,速度快,满足人脸疲劳检测的实时性要求.

Face Fatigue Detection Based on Adaboost Algorithm

The eyes state is the most important and obvious characteristics of the body fatigue.This paper combine color with Adaboost to detect face,on this basis, combine the edge of the face structure characteristics and Adaboost method to position eyes precisely. Then using adaptive threshold and mathematical morphology method to extract the eye feature. At last, integrating PERCLOS rules and nod to judge the fatigue state. Experiments show that this method is robust, fast speed, and meet the real-time demands.