Research On Dynamic LOD Rendering Method Of Vector Map Based On GPU

As a product of map visualization,map serves as a kind of indispensable language and tool in people’s work and life.The interactivity,quality and efficiency of map visualization exert significant influence on the analysis of geographic objects and phenomena and exploration of geographic regularities.With the rapid development and popularity of earth observation technology,sensor and network transmission technology,the scale and variety of spatial data have grown rapidly,which has brought great challenges to map visualization.For the real-time rendering of large-scale spatial data,traditional spatial index technology,which can improve the efficiency acquisition and satisfy the demands of acquiring visual data dynamically from mass data,could hardly solve the problem of long data scheduling time caused by mass data visualization under a small plotting scale.LOD(Level of Detail)can be used as a visual rendering method,which can simplify and render the data under a complex scenario effectively.The static LOD rendering method has superior performance in rendering performance;however,it will generate redundant data storage and require a longer data preprocessing time.At the same time,under adjacent scales,the data visualization results are difficult to smoothly transition.With the dramatic enhancement of graphics processor performance and the development of programmable features,the computer could process graphics better and faster.In the field of GIS(Geographic Information Science),to boost the study on the real-time rendering of large-scale vector datat,the computing performance,graphic display quality and rendering effect of GPU(Graphic Processing Unit)are used fully to transfer the computations such as data coding and subdivision as well as graphic rendering from CPU(Central Processing Unit)to GPU.However,most of the existing researches foucus on map symbol drawing methods.Without considering the computational cost of symbolization brought by massive high-resolution data rendering.Therefore,the high-performance rendering of vector map is still challenging.To address the problems and challenges concerning the vector map rendering method at present,data organization and spatial index structure of vector data were designed for map visualization in this paper so as to study the dynamic LOD rendering method of vector map under a GPU environment.The research contents and results of this paper are as below:(1)Design for data organization and spatial index structure of vector data.First,the map rendering process and GPU rendering pipeline were analyzed,the relationship between map symbol drawing and vector data was summarized.Then the influence of map rendering on data dispatching and data LOD rendering were discussed.Finally,design the data organization and spatial index structure for visualization requirements,combine the hierarchical nature of the spatial index,design the hierarchical coding structure,and realize the high-performance range query and scheduling of vector data in the database.(2)GPU-based dynamic LOD rendering method of vector map.First,analyze map symbolization method of vector data under GPU environment and design the subdivision methods of points,lines and polygons.Then LOD dispatching methods of points,lines and polygons based on the pixel tolerance,data organization and index query structure in this paper.Finally,the dynamic LOD rendering of vector map was realized with the graphics rendering method in the GPU environment.(3)Design of prototype system.Based on the research results above,the prototype system of vector map rendering designed to realize the algorithm proposed in this paper.A contrast experiment concerning the rendering method in this paper,designing high-performance rendering method of current vector map and rendering method of non-tile map to verify the practicability and efficiency of the method in this paper.