中文摘要：

利用基于3DVoronoi多面体分割三维空间,并将其应用于具有典型三维特征的点云数据的聚类分割。通过对点云数据的离散体元表示,透过Voronoi单元的特征参数实现了三维点集的度量、提取和结构分析,揭示了点集间存在的相互关系,并通过3DVoronoi图所确定的空间邻近关系完成点集间相似度的测度和聚类。以三维兔子点云为样本数据的实验分析表明,本文所提出的思路聚类分割特征明显。

英文摘要：

3D point pattern of cluster analysis is presented based on 3D Voronoi. 3D Voronoi cell is used to represent the spatial region that the spatial point effects. Through a quantitative description of the spatial parameters about 3D Voronoi cell there exists the potential to distinguish the weight and effective quantity of each point in the 3D space. Spatial neighborhood relationship among points is extracted according to 3D Voronoi cells to delimit the candidate points that will be clustered. The method illustrates segmentation and cluster distribution of 3D points based on the underlying density and spatial relationships, and actual analysis is imposed on the point cloud of 3D rabbit （Bunny）. The ability to make quantitative description of each 3D Voronoi cell gives insights into spatial controls and cluster process on aD points.