计算机与现代化 ›› 2012, Vol. 1 ›› Issue (200): 179-04.doi: 10.3969/j.issn.1006-2475.2012.04.049

• 图像处理 • 上一篇    下一篇

基于YCbCr肤色模型改进算法及区域特征的人脸检测研究

邱鹏瑞   

  1. 昆明冶金高等专科学校,云南 昆明 650033
  • 收稿日期:2012-03-05 修回日期:1900-01-01 出版日期:2012-04-16 发布日期:2012-04-16

Research on Human Face Detection Based on Improved YCbCr Skin Color Model and Area Character

QIU Peng-rui   

  1. Kunming Metallurgy College, Kunming 650033, China
  • Received:2012-03-05 Revised:1900-01-01 Online:2012-04-16 Published:2012-04-16

摘要: 针对当前人脸检测的研究现状与难题,采用改进的YCbCr椭圆聚类肤色模型进行肤色区域提取,根据肤色在YCb’Cr’空间的分布,对于亮度小于80的非肤色像素点会误判为肤色点,则缩小椭圆聚类; 对于亮度大于230的肤色像素点会误判为非肤色点,则扩大椭圆聚类,有效避免了在高亮度区域和亮度较低的区域中的肤色点误判问题。接着利用人脸的几何特征,对二值化图中的目标区域进行比例、大小结构的分析,排除不可能的人脸区域,并基于肤色和位置进行区域优化,将处理后的结果作为候选人脸区域输出。

关键词: 人脸检测, 肤色模型, 区域特征

Abstract: In accordance with the current study situation and difficult problem of human face detection, the paper adopts improved YCbCr elliptic clustering skin color model to extract color zone from the skin. According to the distribution of the YCb’Cr’, if the color brightness less than 80 pixel of color will be a miscarriage, it narrows the ellipse clustering; if greater than 230 complexion pixel of color will be a miscarriage also, it expands the ellipse clustering. It can avoid wrong judgment about skin color spot between high brightness zone and lower brightness zone effectively. And then taking advantage of the geometric characteristic of human face, it analyzes the objective zone with proportion and size in the binary picture to get rid of the impossible human face.

Key words: face detection, skin color model, area character