双色彩空间综合肤色模型的人脸检测研究

PDF全文下载：2015040449

徐云， 陈志军*， 闫学勤(新疆大学 电气工程学院,新疆 乌鲁木齐 830047)

 摘要：在复杂背景环境下，单一色彩空间和单一肤色模型的肤色分割效果不佳，为了提高肤色区域提取的效果，提出双色彩空间的综合肤色模型的人脸检测方法。即先在HSV(Hue,Saturation,Value)色彩空间用简单肤色模型做粗肤色提取，然后在聚类性较好的YCgCb色彩空间采用综合肤色模型提取法（CCM）进行二次肤色提取，最后再通过SNoW(split up Sparse Network of Winnows)分类器人脸定位算法更加准确地检测到人脸。通过使用改进的方法，得到的检测数据表明，该方法可有效提高人脸检测的检测率。

关键词：人脸检测; 双色彩空间； 综合肤色模型； SNoW

中图分类号： TP 317.4文献标志码： A

 Study of Face Detection Based  on Double Color Space and Comprehensive Skin Color Model

 XU Yun， CHEN Zhijun， YAN Xueqin

(College of Electrical Engineering, Xinjiang University,Urumqi 830047,China )

Abstract: It is not very effective for skin color separation only using single color space and skin color model when the object is in complicated background. In order to improve the performance of skin color extraction, we suggested a new method that using dualcolor model for face detection. The mechanism is that: First crude skin colors are extracted using basic skin color model in HSV（Hue,Saturation,Value）color space. Then skin colors are extracted again by using comprehensive color model in better clustering performanced YCgCb color space. Finally, SNoW(split up Sparse Network of Winnows）classifier using face location algorithm is operated in order to detect human faces more preciously. Overall, after observing test data, we conclude to a result that this method can improve the efficiency of face detection using this advanced method.

Key words:face detection; double color space; comprehensive skin color model; SNoW

收稿日期： 20141205

基金项目： 新疆维吾尔自治区教育厅高校科研计划青年基金项目（XJEDU2012S06）.

作者简介： 徐云(1990—),女, 硕士研究生    *通信联系人.

文章编号：16726987（2015）04044906； DOI： 10.16351/j.16726987.2015.04.019