基于度量直方图的医学图像检索  

Medical image retrieval using metric histogram

作　　者：金华[1] 卢佳[1] 宋余庆[1] 谢从华[1] 

机构地区：[1]江苏大学计算机学院,江苏镇江212013

出　　处：《计算机工程与设计》2005年第11期2952-2954,共3页Computer Engineering and Design

基　　金：镇江市社会发展基金项目(SH2003014);江苏大学高级人才基金项目(04JDG043)

摘　　要：研究了医学图像灰度分布的特性,利用相邻灰度的相关性提出了基于度量直方图的医学图像检索方法及其度量空间上的距离函数MHD();该方法减少了传统直方图特征的维数,克服了SAM索引的缺点。通过对CT图像数据库的检索实验,验证了该方法在性能和速度上都超过了传统直方图检索方法。The gray distribution of image was discussed. A metric histogram for content-based image retrieval and a new metric distance function MHD 0 were presented by taking advantage of the correlation among adjacent bins of histograms. The new method reduced the dimension number of feature vectors obtained from conventional histogram and overcame the flaws of spatial access method （SAM）. A database of CT image was retrieved by the method and results show that it has good performance.

关 键 词：医学图像数据库 基于内容图像检索 直方图 度量空间索引 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]