The Research Of Large Scale Terrain Visualization

With the development of the computer graphics, computation geometry and geographic information system, large-scale terrain real-time visualization has been widely used in military, map navigation, urban planning and other fields. The development of satellite remote sensing technology has made it possible to obtain realistic remote sensing image, so render realistic and massive terrain in real time become an urgent need. But the limited computer hardware result in rendering large scale terrain cannot achieve the requirement. Therefore, it is significant and important to realize the high realistic visualization system based on the massive terrain data.First of all, some kinds of terrain rendering algorithm are compared in this paper, Geo Mip Map algorithm based on large memory capacity, high execution speed and the classic data structure is realized. Terrain is divided into many small regular quads using a quad tree, screen space error is calculated to determine whether its child nodes will be calculated. The algorithm reduces the number of calculation, decreases the computational overhead and increases the rendering efficiency.Secondly, the terrain data organization and schedule strategy are analyzed and researched, to construct the multi-resolution large scale terrain model. Terrain division is used in organizing the digital elevation model(DEM) data and satellite image data to decrease the storage space of terrain data and improve the efficiency of data transmission. In scheduling terrain data, the view volume algorithm is improved, the screen error algorithm based on view point is realized, the cached data strategy is designed and the multithreading technology is introduced to dispatch and render terrain data.Finally, 3D large-scale terrain visualization system is designed based on real-time terrain rendering algorithm, the organization and schedule strategy of terrain data and 3D simulation technology. The evaluation result shows that the visualization system can effectively improve the data operation efficiency, reduce the calculation burden of the CPU by using GPU and quad tree structure and maintain higher interactive frame rates. It meets the need of large scale terrain scene rendering in real time, can effectively test weapons performance and assess the operation scheme.