基于肤色和区域双重匹配的人脸检测

Face Detection Based on Dual Matching of Skin Color and Region

提出了一种在复杂背景的图像中自动检测彩色人脸的方法。这种方法将肤色信息与人脸区域信息相结合。先在YCbCr颜色空间中求出图像中每个像素点属于肤色的隶属度,然后求出每个像素点对于质心的区域隶属度,最后把这两个隶属度进行结合得到属于人脸的隶属度。试验结果证明这种方法能较好地在复杂背景中检测出人脸。

In this paper we propose a face detection algorithm for color images in complex backgrounds. Our method combined the information of skin color with the information of regions. In the YCbCr color space, the method computes the skin color membership of every pixel in the photo. Then it computes the region membership of every pixel which related to the center of mass, and finally combines the two memberships above together to get the face membership. Experimental results demonstrate successful face detection in complex backgrounds.