Quadtree-based Dynamic Lod For Real-time Terrain Visualization

With the development of computer and network technology, the traditional2Dvisualization system can not fulfill people’s requirements, for which more and moreresearches committed to the research on3D visualization system. While, at the sametime, the resolution of geospatial data has been getting very detail which made thesize of data increasing rapidly since aerospace and remote sensing mappingtechnology has made greate achivements. Therefore, the establishment of large-scalevisualization of geographic information system to manage and analyze geographicinformation data has become the hotspot. Three-dimensional terrain real-timerendering technology is one of the core technology and difficulties for visualizationsystem.The scale of terrain data greatly exceeded the capabilities of modern graphicshardware. In order to accommodate complex terrain models, methods for usingquadtree-based level of details (LOD) models and generating a view-dependent meshfrom the LOD model during rendering have been proposed. Via find out the problemsof previous algorithms, the corresponding solution are proposed in this paper, andfinally I implemented large-scale terrain surface mutiresolution approximation basedon dynamic LOD. Here are the main study work of this research:(1)I simplifed the design of quadtree structure and proposed a data strucutre withquadtree and matrix, in which not only surface approximation with a multiple levelsof detail hierarchy but also getting next data could be made easy.(2)A bottom-up algorithm was designed to calculate the attributes of terrainblocks with the result of sub-blocks in this paper, which significantly improves theefficiency of the terrain node evaluation.(3)A simplified mesh refinement algorithm was proposed according to thecharacteristics of quadtree. The refinement criterion which would divide the sub-nodeinto4isosceles right triangles if the sub-node was non-leaf-node, but if it was, it willbe divided into8triangles separately.(4)A node evaluation algorithm was designed and implemented based on terrainroughness. After continuous adjustment the parameters of node evaluation equation inserveral experiments, the system has achieved the best result of terrain gridvisualization.