一种基于颜色统计聚类的医学图像检索技术  

An Image Retrieval Method Based on Color Statistical Clustering

作　　者：陈希超[1] 夏顺仁[1] 

机构地区：[1]浙江大学生物医学工程学系,杭州310027

出　　处：《计算机工程与应用》2004年第5期40-41,54,共3页Computer Engineering and Applications

基　　金：国家自然科学基金的资助(编号:60272029)

摘　　要：基于颜色检索的基本思想是将图像间的距离归结为其颜色直方图间的相似性度量,从而图像检索也就转化为颜色直方图的匹配。目前基于颜色检索的算法主要集中在不同颜色空间进行全局颜色聚类或融合其他可视特征(纹理,颜色空间信息等)联合检索两个方向上。该文在具体的结肠镜图像检索系统研究中,根据医学图像的特点,提出一种在HSV空间的颜色统计聚类的检索方法,取得了良好的检索效果。The main idea of image retrieval base d on color feature is using the similarity between color histograms of images to replace the similarity between images,thus the retrieval of images is transferr ed to the matching between color histograms.The conventional methods about ima ge retrieval based on color feature are mainly focused on two areas:global col or clustering at different color space,or combining with other visual features (texture,space information of color,etc.).In this paper an algorithm based on color statistical clustering at the HSV color space is presented,and the ex perimental results show this new method offers significant advantages.

关 键 词：医学图像检索 颜色统计聚类 HSV颜色空间 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]