利用肤色分割和自适应模版匹配的人脸检测  

Face detection using skin-color segmentation and adaptive template matching

作　　者：王智文[1] 蔡启先[1] 陈劲飙[2] 王乃嵩[2] 

机构地区：[1]广西科技大学(筹)计算机学院,广西柳州545006 [2]广西科技大学(筹)土木建筑工程学院,广西柳州545006

出　　处：《广西工学院学报》2013年第1期1-8,106,共8页Journal of Guangxi University of Technology

基　　金：广西自然科学基金项目(2011GXNSFA018162)资助

摘　　要：针对基于模板匹配人脸检测算法计算量大和基于肤色分割人脸检测算法正确率低、易受类肤色背景影响等缺陷,提出了利用人类肤色分割和自适应模板匹配的人脸检测方法.利用人脸肤色混合高斯模型对人脸进行分割实现迅速检测人脸的目的;采用二维多尺度离散正交小波变换来进行光照补偿,克服亮度对人脸检测的影响;利用自适应模板匹配和二次匹配算法来减少类肤色背景对人脸检测的影响及匹配过程中的计算量.实验结果表明:本文提出的方法能快速检测出人脸,克服亮度和类肤色背景对人脸检测的影响,并提高人脸检测的准确率.In order to overcome the deficiencies that computation of face detection algorithm using template matching is very high and the recognition rate of face detection algorithms using skin-color segmentation is low, and is vulnerable to the impact of background which is similar with skin-color, face detection using skin color segmentation for human and adaptive template matching algorithm is presented in this paper. Human faces are separated from the image by using Gaussian mixture model in order to achieve the purpose of rapidly detecting the face of human face. Two-dimensional multi-scale discrete wavelet transform is used to compensate the illumination and to overcome the influence of brightness. Adaptive template matching and second matching algorithm are used to overcome the effect of the backgrounds on face detection and reduce computation in the process of matching. Experimental results show that proposed method can rapidly detect human faces, overcome the influence of brightness and resembling skin-color background in the course of face detection and improve the accuracy of face detection.

关 键 词：人脸检测 自适应模板匹配 光照补偿 混合高斯模型 二次匹配算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]