A Parallel Generation Method Of Vector Tiles For Multi-Scale Visualization

With the popularization of web map applications,vector tile technology is showing infinite vitality.Vector tiles,as an extension to raster tiles,directly store the spatial and attribute information of vector features,which can satisfy the dynamic interaction requirements of users.The generation of multi-scale vector tiles needs to be constructed based on tile pyramids,and as the build level increases,the number of tiles to be built will increase exponentially.In addition,the tile construction process involves a lot of computation intensive and IO-intensive processing operations,these time-consuming operations will seriously affect the overall efficiency of vector tiles generation.Therefore,how to generate vector tiles suitable for multi-scale visualization efficiently is a challenge that worth of studying in the development of vector tile technology.The purpose of this paper is to design and implement a parallel generation method of vector tiles for multi-scale visualization.The method optimizes the slicing performance from several aspects such as spatial data organization and indexing,tile format,parallel program design,and etc.At the same time,slicing and multi-scale representation are achieved by improving general data processing and generalization operators.The research in this paper can be summarized as follows:(1)This paper proposes both efficient spatial data organization and query methods for vector tiles.Based on the organization characteristics of vector tiles,and combined with highly balanced STR R-tree indexes,a hybrid index structure that adapts to different morphological distributions of vector data is proposed,which optimizes the vector data source retrieval method and speeds up the generation of vector tiles.(2)This paper designs and implements a vector slicing method that suits to the multi-scale visualization of maps.For multi-scale scaling of vector maps,multi-scale representation strategies for vector data in different application scenarios are proposed,and feature-oriented vector data reduction and selection methods are implemented,which avoid the loss of topological relationships when large-scale vector data is displayed in multiple scales.(3)Optimizes vector tile clipping method based on tile rendering mechanism of browsers.Based on the visualization mechanism of the Canvas container,a four-to-extended vector tile clipping method is implemented,which solved the problem of visible cut lines after the tiles were merged,so the correct merging of vector tiles on the visualization level is realized.(4)A vector tile parallel slicing scheme in stand-alone and distributed environments is proposed.Firstly,the feasibility of parallelizing slices is analyzed based on hardware and software and slice model support.Secondly,based on the general parallel computing framework,the parallel generation of vector tiles is realized.Finally,the performance comparison experiment verifies the effectiveness of the parallelization scheme.