计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (11): 260-265.DOI: 10.3778/j.issn.1002-8331.2003-0172

• 工程与应用 • 上一篇    下一篇

改进核密度估计的空间点密度算法

饶加旺,马荣华   

  1. 1.江苏省测绘工程院 空间信息技术研究中心,南京 210013
    2.中国科学院 南京地理与湖泊研究所,南京 210008
  • 出版日期:2021-06-01 发布日期:2021-05-31

Improved Kernel Density Estimator Based Spatial Point Density Algorithm

RAO Jiawang, MA Ronghua   

  1. 1.Spatial Information Technology Research Center, Jiangsu Province Surveying & Mapping Engineering Institute, Nanjing 210013, China
    2.Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 210008, China
  • Online:2021-06-01 Published:2021-05-31

摘要:

针对常用的核密度估计在计算离散点密度上存在的诸多局限,提出了空间点密度算法。该算法在保持点离散属性与初始空间位置的基础上,设定分箱规则,获取离散点最近的格网点坐标;通过遍历,统计并计算每个搜索邻域内点的数量,以离散点初始坐标与点密度值为输出结果。以USGS的美国大陆地下水资源数据集展开实验研究,采用可视化输出与时间复杂度为验证指标,与核密度估计算法进行对比验证。实验结果表明,该算法提高了点密度的识别性,获取了离散点真实的密度值,可视化效果与精度方面均优于ArcGIS 10.4.1与kde2d核密度分析的结果、运算效率优于kde2d算法。

关键词: 地理空间, 点密度, 离散点, 哈希表, 数据挖掘

Abstract:

Due to the negative effects of commonly used kernel density estimator has many limitations on computing discrete point density, a point density algorithm based on projection and Hash-based data structure is purposed, which can quickly compute the true density of discrete points, by keeping the original location and discreteness of points. Coordinate of points in the nearest neighborhood grid are obtained by the rule of binning, then it calculates the number of points in each search neighborhood, density value and initial coordinate with discrete points for the output results. Groundwater resources of the continental United States extracted by USGS are used as experiment dataset, visualization output and the time complexity are used for validation indexes when compared with algorithm based on the kernel density. Results show that the algorithm improves the identification of density, obtains the true density value of discrete points, visual effect and the precision analysis are better than that of ArcGIS 10.4.1 kernel density and kde2d kernel density algorithm, the computation efficiency is better than common kde2d algorithm.

Key words: geographic spatial, point density, discrete points, Hash table, data mining