基于肤色分割的人脸检测算法

Face Detection Algorithm Based on Skin Color Segmentation

研究一种基于肤色分割的人脸检测算法。首先将图像色彩空间由RGB转化为YCbCr;其次根据肤色点在色彩空间的聚类性能建立肤色区域模型和简单高斯模型;最后通过计算肤色相似度得到肤色似然图,经过平滑滤波、形态学处理和阈值分割等图像处理过程,确定人脸位置。实验结果表明,该算法能够较好地实现人脸检测。

A face detection algorithm based on skin color segmentation is studied.Firstly,the color space of image is transformed from RGB to YCbCr;secondly,the skin color region model and simple Gaussian model are established according to the clustering performance of skin color points in color space;finally,the skin color likelihood map is obtained by calculating skin color similarity,and the face position is determined through image processing such as smooth filtering,morphological processing and threshold segmentation.The experimental results show that the algorithm can achieve face detection well.