一种新的肤色相似度图自适应阈值选取方法  

A NOVEL SELF-ADAPTIVE THRESHOLD CHOSEN METHOD OF COMPLEXION SIMILARITY IMAGE

作　　者：曾宪贵[1] 左文明[2] 石玉强[1] 刘磊安[1] 

机构地区：[1]仲恺农业工程学院计算机科学与工程学院,广东广州510225 [2]华南理工大学经济与贸易学院,广东广州510006

出　　处：《计算机应用与软件》2010年第9期125-127,共3页Computer Applications and Software

基　　金：广东省软科学研究(2007B060401078)

摘　　要：针对复杂背景图像,基于肤色相似度直方图最低波谷自适应阈值选取法分割提取肤色区域效果不理想,提出第一波谷自适应阈值选取法。实验结果表明,该方法对复杂背景图像的肤色区域分割提取效果较为理想。For images with complex background, the effects of segmenting and extracting complexion region is unsatisfactory when using self-adaptive threshold chosen method based on lowest wave trough of histogram of complexion similarity. Hence a novel method of first wave trough self-adaptive threshold is presented. Experimental results indicate that it has an ideal effect in segmenting and extracting complexion region of images with complicated background.

关 键 词：自适应阈值 肤色区域分割 肤色区域检测 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]