摘要
本文通过对图幅进行网格化划分,建立网格索引,对弧段按网格建立R树空间索引,进一步降低了空的相交判断次数;根据网格之间一定的独立性,对不同网格内弧段进行并行化求交,并在单CPU多核计算机上利用OpenMP并行机制对算法进行了实现。分析与实验表明:改进后的新算法对较大数据量弧段求交的处理效率较高,与同类算法相比,在空间数据拓扑的建立与空间分析的应用中具有一定优势。
A parallel algorithm for arc index with gridding thought. It aims at int intersection was proposed based on grid index and R-tree spatial ersectlon terseeting judgment were further decreased by p grids. According to the independency of grids, the tion. The algorithm was realized on a single-CPU grammmg rithm coul algorithms rithm wou issues of large amount spatial data. Empty times of in- artitioning map and building R-tree spatial index to algorithm parallelly processes different grids intersec- and multi-core computer based on multi-threaded pro- thought and OpenMP mechanism. The analysis and experiment showed that the improved algo- d process a large amount of intersection for arc with high precision. Compared with the similar in the applications of establishing topology and spatial analysis on spatial data, the novel algo- ld be superior.
出处
《测绘科学》
CSCD
北大核心
2014年第3期111-115,共5页
Science of Surveying and Mapping
基金
国家863基金资助项目(2009AA121404)
关键词
弧段求交
网格索引
R树
并行算法
intersection for arc
grid index
R-tree index
parallel algorithm