基于HSV和纹理特征的相容near度量方法  

Tolerance Nearness Measure Based on HSV and Texture Feature

作　　者：刘文英[1] 王拥军[1] 杨义川[1] 

机构地区：[1]北京航空航天大学数学与系统科学学院,北京100191

出　　处：《计算机科学》2015年第B11期109-112,134,共5页Computer Science

基　　金：国家自然科学基金(11271040)资助

摘　　要：基于内容的图像检索是图像处理研究的重点,而相似性度量是其核心问题。基于near集的tNM(Tolerance Nearness Measure)方法在仅提取图像的灰度值特征时比IRM(Integrated Region Matching)检索结果更好。基于tNM与人类视觉近似的特点,将灰度值替换为面向用户视觉的HSV(Hue,Saturation,Value)颜色空间,分别提取图像的灰度值(Grey)+纹理(Texture)、HSV+纹理两组特征。使用IRM和tNM算法对10类图像进行检索,对其检索结果进行比较分析,结果表明使用tNM算法提取的图像的HSV+纹理特征与人类视觉更加近似,效果更佳。Content-based image retrieval is a very important issue in image processing. Similarity measure is a core prob lem in content-based image retrieval. Tolerance nearness measure method based on near set is better than IRM（integra- ted region matching） when it just extracts the Grey feature. Considering that tNM is close to human visual, we replaced Grey feature with HSV color space. We extracted Grey＋texture feature and HSV＋texture feature, respectively. Then the retrieval results was obtained by IRM and tNM from 10 categories images. Through analyzing and comparing those results, we drew a conclusion that HSV＋ texture feature has higher performance compared to Grey＋ texture feature.

关 键 词：相似性度量 相容near度量 灰度值+纹理 HSV+纹理 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]