基于肤色模型和中线定位的人脸检测算法

由于外貌、肤色、表情等不同，会导致较高的人脸检测漏检率和误检率。为此，提出一种基于肤色模型和中线定位的多姿态人脸检测算法。利用肤色特征快速排除大部分背景区域，根据人脸的显性特征分割出人脸候选区域，并对边缘检测后的图像进行投影，使用中线定位法实现多姿态人脸的检测与定位。实验结果表明，该算法能实现多姿态人脸的快速检测，黑发单个人脸检测的检测率达93.3%，鲁棒性较强。

Face Detection Algorithm Based on Skin Color Model and Midline Localization

The appearances of human face, expression, color and so on are different, which bring difficulties to face detection, such as high false face detection and false dismissal. This paper proposes a multi-face detection algorithm based on skin color model and midline localization. Utilizing the facial skin color information firstly, most of the background regions can be excluded quickly. The approximate face candidate regions can be segmented according to facial external structure features. By projecting the edge detection image to X-axes, the midline position of face can be located, and multi-face detection and localization are realized. A black hair single face detection rate of 93.3% can be obtained for a test photo sequence. Experimental results show that the algorithm proposed in the paper can improve the multi-face detection accuracy and is highly robust to facial expression.