| With the rapid growth of massive spatial data,spatial clustering algorithm becomes one of the important basic means to mine spatial value information.At present,the massive spatial elements present complex spatial characteristics,and the elements under different spatial scales have different distribution states.Therefore,spatial clustering should have spatial scale dependence,and the clustering results under different spatial scales are different.Especially in the current widespread use of Internet maps,the scaling function of multiple spatial scales has matured,which makes the rapid display of multi-scale spatial aggregation elements a new requirement of the current clustering method.In view of the low accuracy of clustering algorithms provided by Internet map vendors and the difficulty of traditional clustering algorithms to quickly process massive element data at multiple spatial scales,this paper aims to study the massive element clustering algorithm adapted to the WEB terminal at multiple spatial scales,improve the clustering efficiency,realize the fast processing of large-scale data,and meet the fast clustering and display of elements at different visual scales.Therefore,this paper proposes a method called multi-scale massive points fast clustering based on hierarchical density spanning tree.On the basis of fully understanding and reproducing various types of clustering algorithms,the characteristics and shortcomings of each algorithm are summarized;by referring to the basic principle of CFSFDP algorithm,the concept of hierarchical density spanning tree based on hierarchical density is introduced,and the difference between the tree structure of hierarchical density tree and other clustering algorithms;the corresponding pruning strategy is proposed in combination with the spatial scale and the tree connection of elements,and the multiscale fast clustering of elements is finally realized.The experiment is divided into two major blocks.The first part is the multiscale aggregation of point elements.Firstly,we use the simulated point data to verify its clustering effect and illustrate the different density scenarios that different densities can cope with;secondly,we use the POI and landslide hazard point data to prove its superiority in clustering speed and accuracy of results at different spatial scales,and show the corresponding clustering results to illustrate the clustering characteristics of MSCHT and the role of hierarchical density generation tree.The second part is the expansion of clustering of polygon elements,and the MSCHT clustering method is implemented by using multi-density attributes to calculate the indexes for the current problems of different sizes,different forms and uneven distribution of front elements,which lead to less accurate clustering results.At the same time,the algorithm in this paper has good accuracy performance in both simulated data and experimental results of real data,and meets the demand of aggregation processing for polygons on the basis of maintaining the multi-spatial scale characteristics. |
