基于网络加权Voronoi图的点群选取

传统基于Voronoi图的算法忽略了点与点之间是通过实际网络距离相连这一事实，针对此缺陷，提出一种基于网络加权Voronoi图的点群选取算法. 1）利用网络扩展法构建点群的网络加权Voronoi图；2）计算每个点对应的网络Voronoi多边形面积及扩展弧段总长度，并以此为依据，为点群中所包含的统计、专题、拓扑和度量信息分别选定量化描述因子；3）提出“同心圆”算法，解决点群取舍问题. 实验结果表明，提出的方法顾及了点群权重以及与点群相关联的道路等级、方向及局部密度对选取结果的影响，较好地保持了原始点群的各类信息，选取结果符合实际地理空间特征.

Point cluster selection based on weighted network Voronoi diagram

The existing algorithms based on Voronoi ignores the fact that points are connected through road network, in view of which, a new algorithm was proposed based on network weighted Voronoi diagram. Firstly, the weighted network Voronoi diagram of the point cluster was constructed based on expansion operation. Secondly, the area of the weighted network Voronoi polygon and the total length of the expansion route of each point were calculated, based on which the appropriate factors for describing the statistical, thematic, topological and metric information were selected. Thirdly, the method called ‘concentric circle’ was proposed and the point deletion was completed. The experimental results show that the algorithm takes into account the effect of the weight of the point, the level and the direction of the roads, and the density of the road network on the generalized results. The information of the original point cluster is transmitted well and the generalized results fit the actual geographic feature.