基于YCgCr颜色空间的肤色检测方法  

Algorithm of skin detection based on color space YCgCr

作　　者：康进峰[1] 王国营[1] 

机构地区：[1]解放军信息工程大学电子技术学院,河南郑州450004

出　　处：《计算机工程与设计》2009年第19期4443-4446,共4页Computer Engineering and Design

摘　　要：提出了一种可变光照条件下的肤色检测算法。该方法分析了YCgCr颜色空间特征,使用贝叶斯规则来建立肤色分类器,在Y-Cg和Y-Cr两个子空间联合建立完整的肤色模型,分析比较了在相同方法下YCbCr和YCgCr颜色空间的优越性。实验结果表明,YCgCr颜色空间上得到的误检率和漏检率均低于YCbCr颜色空间,该算法可适用于较广范围的光照条件,可以有效地从复杂背景的彩色图像中检测出肤色区域。A skin color detecting algorithm under variant light condition is presented. Firstly, the feature of YCgCr color space system is analyzed, and a skin color classification technique is adopted that employs Bayesian decision with color statistics data, the skin color model is constructed by two lookup tables in Y-Cg and Y-Cr subspaces. After comparison using the same method it＇ s concluded YCgCr color space is better than YCbCr color space. Experimental results have demonstrated that false detection rate and false dismissal rate at YCgCr color space are both lower than at YCbCr color space, this method can fit under wider light condition, and it＇ s also effective to detect skin-color regions from the color images with complex background.

关 键 词：YCGCR颜色空间 查询表 贝叶斯规则 肤色检测 错误率 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]