基于内容图像检索中的一种动态多维索引方法  

A Dynamic Multi-dimension Index Method in Content-based Image Retrieval

作　　者：徐焕[1] 林坤辉[1] 周昌乐[2] 

机构地区：[1]厦门大学软件学院,福建厦门361005 [2]厦门大学信息科学与技术学院,福建厦门361005

出　　处：《计算机工程与应用》2006年第23期161-164,共4页Computer Engineering and Applications

基　　金：国家985一般项目(编号:9850010)

摘　　要：多维索引技术是基于内容检索的图像数据库的关键技术。SR-tree和X-tree是目前比较成熟有效的多维索引技术。为了提高多维索引的性能,我们在分析SR-tree和X-tree的结构和性能的基础上,针对SR-tree分裂算法的不足,引入X-tree中超级节点的思想,通过改进插入和分裂算法,设计了一种新的多维索引结构ESR-tree,即ExtendedSR-tree。实验表明,随着数据量和维数的增多,ESR-tree的性能明显优于SR-tree和X-tree。Multidimensional indexing technology is the key technology of content-based retrieval in image database.SR-tree and X-tree are riper and more effective multidimensional indexing technology at present.In order to improve the performance of the indexing,we assay the structure and capability of SR-tree and X-tree.In view of the deficiencies of split algorithm in SR-tree,the paper designs a new multidimensional indexing structure ESR-Tree,that is Extended SR-tree,which introduces the idea of super node of X-tree and improves the insert and split algorithm.The results of the experimentation show that the capability of ESR-Tree is better than SR-tree and X-tree with the increasing data number and dimensions.

关 键 词：基于内容检索 多维索引 超球体 超矩形 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]