基于相似性度量方法的图像检索  

SIMILARITY MEASURE-BASED IMAGE RETRIEVAL

作　　者：宋艳[1] 刘方爱[1] 

机构地区：[1]山东师范大学信息科学与工程学院,山东济南250014

出　　处：《计算机应用与软件》2007年第10期166-168,共3页Computer Applications and Software

摘　　要：图像检索的目的是找出检索对象集中的所有与指定样本图像相似的图像。基于内容的图像检索近年来得到了广泛的研究,人们已经提出了许多基于特征的图像检索算法,在以往的算法中表达图像特征的相似度通常采用距离法,但是这种距离法存在很多不足之处。为了克服这些不足,介绍了一种特征矩阵的构造方法并利用特征矩阵来计算图像的相似度,以此来进行图像检索。Image retrieval is aimed at discovering the similarity between retrieval object centralism and assigned sample images. In recent years, content-based image retrieval has been studied widely, and people have proposed many methods about image retrieval. In order to describe the similarity, distance is adopted. But this method has many disadvantages, and in order to overcome these disadvantages, Feature Matrix is introduced and used to compute similarity and do image retrieval.

关 键 词：基于内容的图像检索 特征矩阵 动态规划 相似性度量 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]