基于Haar-like和肤色特征的驾驶员脸部检测融合算法

对驾驶员面部疲劳状态进行视觉监测的前提是脸部区域的准确、快速检测。采用改进的基于Haar-like特征的人脸检测算法检测出可能存在的初始人脸区域，然后适当扩大初始人脸区域范围，并在此基础上利用肤色特征和区域连通算法在YCbCr和rgb颜色空间上对人脸区域进行二次定位，最后根据定义的脸部区域重合度和人脸几何特征，实现脸部区域的融合检测。实验结果验证了该算法的准确性和可靠性。

Fusion algorithm of driver's face detection based on Haar-like and skin color features

Accurate and fast face detection of driver is the precondition for monitoring driver＇s face fatigue status by machine vision. An improved face detection algorithm based on Haar-like feature was used to detect the face area possibly existing in the image. Then the detected area was extended properly and an face detection algorithm based on skin color feature and region connecting in YCbCr and rgb space was used to locate the face again in the extended area. Finally, accurate fusion detection of face was achieved according to the defined area coincidence degree and the geometric features of human face. Experiment results demonstrated the accuracy and reliability of the algorithm.