基于颜色重心和分层过滤结构的人脸检测算法

提出一种基于颜色重心六边形（colorbarycentershexagon，CBH）模型，结合分层过滤结构的人脸检测方法。首先根据肤色在空间的聚类特性检测人脸，利用CBH模型将彩色图像中的肤色区域和非肤色区域分开，得到人脸检测的候选区域；然后利用级联型的分层过滤结构，采用最优椭圆拟合、形态及方差验证、方向归一化和模板匹配等操作逐步排除非人脸区域，最终确定并标记出彩色图像中人脸的位置。实验结果表明：该方法能够在复杂背景的彩色图像中快速并且准确地检测出人脸，误检率较低，具有较高的应用价值。

Face Detection Using Color Barycenter and Hierarchical Filtering Structure

This paper introduces a method based on color barycenter hexagon model and hierarchical filtering structure which detects faces in color image. In the first step, according to the clustering charac- teristics of skin color in the space, the method separates the skin regions from non-skin regions using CBH model, and gets the face candidate region. Then in order to eliminate the non-human faces, the authors organize some operations on a cascade structure which includes best-fit ellipse, shape and variance verifications, orientation normalization and template matching. At last, it identifies and marks the location of the human face. The results of experiments show that the method can have rapid and accurate detection of a human face in the color image with complex background, the rate of false detection is low, and has a great value in application.