一种空间分布模式驱动的空间索引

支持批量操作的空间索引中,空间数据的分解粒度、局部更新操作的整体影响处理是两个主要难点。本文基于空间分布模式分析,提出了一种空间索引——Pattern-tree。针对批量操作的粒度问题,设计了一种基于空间分布模式探测的空间划分方法,采用一种自上而下与自下而上相结合的索引树构建算法;针对局部插入操作对索引树的整体影响与索引树的调整问题,提出了一种基于空间分布模式变化检测的索引更新方法。试验表明,本文所提出的空间索引结构比STLT、GBI以及SCB等方法具有更高的构建与窗口查询效率。

A Spati al Distribution Pattern-driven Spati al Index

Packing spatial data into blocks and processing of global impact of local operations are two important tasks for spati al index to support bulk operations .In this paper ,we present a new spati al index called Pattern‐tree for bulk operations with spatial distribution pattern analysis .For packing objects into blocks ,a new spati al data partitioning method based on the detection of the spati al distribution pattern was presented .This paper introduces a novel spati al index construction algorithm that combines of top‐down and bottom‐up methods;For processing of local update operations and its global impact ,this paper introduces a new algorithm based on change analysis of the spati al distribution pattern .Empirical results demonstrate that performance improvements are achieved in practice in the case of spatial index construc‐tion and windows query compared with STLT ,GBI and SCB .