Recognition,Measurement and Generalization for Point Cluster Features in Digital Nautical Chart

This paper,based on the Delaunay triangulation and Voronoi diagram model,focuses on the discussion of spatial distribution properties by recognition and measurement.Four characteristic parameters are defined for distribution property description:distribution density of three dimensions,distribution range of two dimensions,distribution axis of one dimension,distribution center of zero dimension.With the aid of Delaunay triangulation and Voronoi diagram,description and calculation models of above-mentioned parameters are established. Firstly,considering the visual principles fully,a new method,which finds the distribution polygon range by "nibbling the outside triangles",is presented.Furthermore,different results can be gained by using different threshold values,so the continuous-scale display may become real. Secondly,the distribution density is represented by Voronoi cell size and visualized as gray image.So the density can be changed into the area that can be defined and measured easily.We may know where is denser and where should be simplified firstly.But one thing must be pointed out here that every point is regarded as nonobjective,and they divide the space on the equal principle.It is the basis and accords with the Voronoi principle. Thirdly,the distribution center can be extracted from gray image.The new concept and methods mentioned above are integrated into a recognition and measurement model for spatial distribution properties of point cluster.It becomes the basic of point cluster generalization. Finally,a generalization model of point cluster is provided in this paper on the basis of Voronoi diagram establishment in a dynamic way.According to the principle,an iterative method is proposed.Through different extractive times,different scale span results can be obtained. In a word,the generalization model,which is established in this paper,preserves the spatial distribution properties of point cluster very well.And all the methods mentioned above have been proved to be feasible and sound.