一种彩色人头图像面部特征区域快速提取方法  

A Quick Extracting Method of Facial Feature Region of Color Head Image

作　　者：陈宇拓[1] 韩旭里[1] 余英林[2] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]华南理工大学电子与通信工程系,广州510641

出　　处：《现代计算机》2007年第4期4-7,共4页Modern Computer

基　　金：国家自然科学基金项目(60372068)

摘　　要：利用彩色人头图像在RGB色彩空间三个色彩分量之间存在的密切相关性,提出只对一个色彩分量进行分析提取,来确定彩色人头图像的面部特征区域,提取过程采用积分投影图和局部特征扫描分析,有针对性地选择边缘检测算子确定局部轮廓。该方法算法复杂度低,实验结果表明能得到较精确和满意的提取结果。This paper presents a method to determine the facial feature region of a color head image by analyzing and extracting the facial feature region from only one color component using the close correlation existing between three color components of the color head image. In extracting process, the analysis of integral projection graph and partial feature scan were used, and the arithmetic operators of edge detection was accordingly selected to confirm partial outline. This method is of lower complexity of the algorithm, and experiment results show that it can obtain satisfactory and precise extraction effect.

关 键 词：面部特征 积分投影 局部扫描 边缘检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]