基于颜色空间分布熵的图像检索  

Image Retrieval Based on Color Spatial Distribution Entropy

作　　者：孙全党[1] 高美真[2] 

机构地区：[1]河南师范大学计算机与信息技术学院,河南新乡453007 [2]焦作师范高等专科学校计算机与信息工程系,河南焦作454001

出　　处：《微电子学与计算机》2009年第10期134-136,共3页Microelectronics & Computer

基　　金：国家自然基金项目(60775045)

摘　　要：颜色直方图是基于内容的图像检索(CBIR)中的一种重要特征,然而其完全丢失了图像颜色的空间分布信息.文中采用颜色空间熵描述图像颜色的空间分布特征,为了消除熵的对称性对图像检索的影响,提出了改进算法.通过实验比较表明了该方法在进行图像检索时是有效的,并具有较高的检索性能.Color histogram is an important technique for color image database indexing and retrieving. However, the main problem of color histogram indexing is that it does not take the spatial relationship of the colors into consideration. We use the color spatial distribution entropy to describe the global spatial relationship of colors. In order to eliminate the effect that the symmetry of entropy has on image retrieval and to further improve the retrieval ability of algorithm, According to the analysis and comparison, this kind of algorithm has a good retrieval ability of image.

关 键 词：图像检索 环形颜色直方图 熵的对称性 颜色空间分布熵 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]