基于颜色块的半径和角度直方图的图像检索  

Image Retrieval Based on Radius and Angle Histograms of Color Blocks

作　　者：黄晶[1] 倪林[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系,合肥230027

出　　处：《计算机工程》2008年第10期202-204,234,共4页Computer Engineering

基　　金：国家自然科学基金资助项目(6037059;60672055)

摘　　要：提出一种基于颜色块的半径直方图和角度直方图的图像检索方法。该方法将图像内容看成由若干颜色块组成的集合,根据每一块颜色块的半径直方图和角度直方图来计算图像间内容的相似度,并进行检索。对Corel图像库的10类图像进行测试,每类取20幅作为查询事例。实验结果表明,该方法与全局直方图、结构量化直方图相比,其平均查准率分别提高了10.3%,11.9%。A new image retrieval method based on Radius and Angle Histograms of Color Blocks（RAHCB） is proposed. The method views image content as a set of several color blocks. The radius histogram and angle histogram of each color blocks are used to compute the similarity of images for image retrieval. Ten groups of images from Corel database are tested. It selects 20 widths of images as inquiry examples. Results show that lhe average rates of increase precision of RAHCB compared with color histogram and structure quantification histogram are 10.3% and 11.9% respectively.

关 键 词：彩色图像检索 颜色块 半径直方图 角度直方图 结构量化直方图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]