基于R-tree的多重近似索引研究

Research on multi-approximate index based on R-tree

本文提出了一种新的索引方法——基于R-tree的多重近似索引,该索引结构既包含实体传统的外部近似也包含其内部近似。在检索时,粗过滤阶段测试实体内部和外部近似,如果内部近似测试为真值就不需要进行详查阶段的处理,从而减少访问磁盘I/O次数而提高检索效率。通过实验验证分析,该索引方法能极大地提高空间数据的检索效率。

A new index method, multi-approximate index based on R-tree was proposed in this paper. It contains both the MBR and the maximum enclosed circle （MEC） of spatial object. Each query candidate before refinement step is tested by exterior and interior approximation. The number of disk accesses in spatial query will be reduced because the candidate is one of the query results if interior test to be true. A series of tests of the multi-approximate index based on R-tree was presented, which indicates that this new method can improve spatial data retrieval efficiency greatly.