摘要
根据公路工程地理数据的空间和属性特征,建立了一种倒排网格索引,通过坐标来反映空间对象在网格中的具体位置。重点探讨了k NN查询算法,对串行轮圈访问k NN算法进行了改进,打破了轮圈半径对其上一次遍历结果的依赖性,以网格边长递增的方式更新轮圈半径,并结合多线程技术实现了多个轮圈的并行访问。通过在模拟的海量公路空间数据集上的实验,从数据集规模、网格边长、k值选取等方面对比分析了两种算法的查询效率。结果表明,改进后的k NN算法对于大规模空间数据集的查询效率有很大提高。
Based on the spatial and attribute characteristics of highway geographical data, we established an inverted grid index, which reflected the spatial location of the spatial object in the grid by coordinates. And then, we discussed and improved the k NN query algorithm. The dependence of rim radius on the previous traversal results was broken, and the radius of the rim was updated by increasing the grid edge length. We combined the multithreading technology to realize the parallel access of multiple rims. The experimental result shows that the improved k NN algorithm can improve the query efficiency of the scale spatial data set.
出处
《地理空间信息》
2018年第5期35-37,40,共4页
Geospatial Information
基金
交通运输部2014年度科技资助项目(2014364J03090)
