基于YCgCb新颜色空间的人脸检测算法的研究  

Research on face detection algorithm based on a novel color space YCgCb

作　　者：郭秀梅[1] 赵秀艳[1] 王玉亮[1] 杨峰[1] 刘贤喜[1] 

机构地区：[1]山东农业大学机械与电子工程学院,泰安271018

出　　处：《电子测量技术》2008年第4期79-81,共3页Electronic Measurement Technology

摘　　要：人脸检测是计算机视觉、模式识别、人机交互研究中一个很受关注的研究热点。人脸模式错综复杂、易受干扰,常见的人脸检测算法一般都具有计算量大、速度慢、误报率高的弱点。本文基于一种新的色度空间YCgCb,建立了亮度和Cg、Cb色度查找表联合的肤色模型,提出了一种快速有效的阈值估计方法。实验表明该肤色模型可以有效地检测到人脸位置,误检率低。Face detection is a hot research of computer vision,pattern identification and human-computer interaction. Face model is anfractuosity and been disturbed easily, the familiar face detection algorithm has a lot of shotcomin＇gs, such as large calculate.low speed and high error detection. This paper constructs a combine model using a table combined by light（Y） and chroma（Cg.Cb）, Put forward a kind of fast and valid a method of estimating threshold value. The experiment expresses that complexion model can examine a person face position availably,and the rate of mistake check is low.

关 键 词：人脸检测 YCgCb颜色空间 肤色建模 阈值分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]