基于肤色的实时人脸检测算法研究  

Research of real-time face detection algorithm based on skin color

作　　者：翟懿奎[1] 黄聿[1] 谢剑华[1] 吴志龙[1] 郭承刚[1] 

机构地区：[1]五邑大学信息工程学院,广东江门529020

出　　处：《电子设计工程》2011年第9期37-39,共3页Electronic Design Engineering

基　　金：国家自然科学基金项目(61072127);广东省自然科学基金项目(10152902001000002;07010869);广东省高等学校高层次人才项目(粤教师函【2010】79号);广东省大学生创新实验项目(1134910031)

摘　　要：研究了一种基于肤色的人脸检测算法的设计与实现过程。在YIQ颜色空间中,进行了有效的肤色提取,把提取到的肤色与背景图像信息转为二值图像进行形态学降噪处理,再采用质心定位法进行准确的眼睛定位,最后对检测到的人脸图像进行缩放、旋转、移位校正,以提高输出人脸图像的质量。实验结果表明,本文所提的算法具有很高的准确率与检测速度,能够适用于实时人脸检测系统。Design and implementation process of a color-based face detection algorithm was studied in this paper. The effective skin color was extracted in the YIQ color space, then the extract skin color and background image information was converted to the binary image and denoised by the morphology processing, the centroid localizing method was used to localize the accurate position of double eyes, finally the scaling, rotation, shifting and rectification operation of the detected face was process in order to improve the quality of the output detection. Experimental results show that the proposed algorithm has high accuracy and detection speed, can be applied to real-time face detection system.

关 键 词：人脸肤色 质心定位法 实时人脸检测 人脸校正 

分 类 号：TP914.53[自动化与计算机技术]