Point Cluster Generalization Approaches Taking Into Account The Weights Of The Points

Many objects are represented as point symbols on the map.Point symbols will be encounter or overlapped with each other and the map becomes illegible when the scale of a map containing point symbols decreases.To keep the clarity,level feature and aesthetics of the map,the point cluster will be simplified,this process is so-called Point cluster generalization.Point cluster generalization means retain important points and delete less important points,through which the information of the original point cluster can be retained with the decrease of the number of the points.Thus,the weight of the point,that is a key factor in point cluster generalization,should be considered in the generalization because it reflects the importance degree of a single point in the point cluster.Throughout the development of the map generalization,it can be concluded that the existing algorithms can be divided into two categories.One type of the algorithms ignores the weight of the point in generalization.The other type of the algorithms takes into account the weight of the point in the generalization,but it has some defects such as the setting of the weight of the point is lack of science,the weight of the point can't reflect the latest information of the point and the road network is ignored in importance evaluation of the point.Aiming at these problems,the influencing factors of the weight of the point are studied and calculated,based on which the point clusters are divided into three different types first.Then,the technologies and theories of corresponding data,pulse coupled neural network,weighted network Voronoi diagram and the Delaunay triangulation are introduced into the research of the point cluster generalization.The major achievement of the thesis can be included as five aspect as follows:(1)The influenced region and the number of the influenced population are considered as the basis of the weight of the point.The value of the area of the influenced region and the number of the influenced population are obtained by means of the technology of the big data,based on which the point cluster generalization is done.The algorithm makes up for the shortcomings of the existing algorithms(i.e.the weight of the point can't reflect the current situation,the algorithms don't take into account the effect of the influenced population to the importance of the point).(2)The weighted network Voronoi diagram is introduced in order to take into account the influence and constraint of the road network to the point cluster.The improved pulse coupled neural network is introduced and the weighted Voronoi diagram is constructed using the principle of the concurrency and automatic wave.The weighted network Voronoi diagram provides a technical reference for the further research on the point cluster generalization and other relative researches.(3)Based on the weighted network Voronoi diagram,network Voronoi polygons are constructed by means of Delaunay triangulation and “stripping” by dynamical thresholds.The area of the network Voronoi polygon and the total length of the road segments in the polygon are considered as the basis for the weight of the point,the proposed method of “concentric circles” is used to realize the generalization of the second type of the point.The drawbacks in the existing algorithm(i.e.points are connected by straight line and the effect of the road network to the importance of the point is ignored)are made up by the proposed algorithm.(4)When it comes to the points whose weights cannot be measured,the algorithm chooses to maintain its contour and geometric structure features to the great extent.The contour points of the point cluster are extracted by methods of Delaunay triangulation and “stripping” by dynamic threshold.Based on which the contour points are simplified by the method of DouglasPeucker and the inner point are simplified by the Voronoi diagram constrained by the contour of the point cluster.The algorithm makes up for the shortcoming of the existing algorithms(i.e.the contour points and the inner points are simplified respectively,which doesn't take into account the interactions between them)Finally,experiments are done to testify the efficiency and soundness of the proposed algorithms.Results show that the weights of the points are considered and other information are transmitted well,the algorithms made up the shortage of the existing algorithms such as and can be used to solve the problems of the generalization of various point clusters.