Research On Large-Scale Terrain Real-Time Visualization Base On Quadtreee And Blocking Technology

Terrain information visualization plays a very important role in real-time emulation and the visual geographical information system. With the constant deepening and the expansion of its application, especially after putting forward "digital earth", more requirements are brought forward about the real-time visualization of 3D terrain. People hope to observe more vivid and widely-ranged terrain in a more natural way.Theoretically speaking, the complexity of the scene is limitless, and the treatment capability of any hardware in time of unit is always limited. Therefore, by investigating the internal relationship between the scene complexity and the relevant factors, such as laws of human visual perception, it is necessary to work out the fast drawing algorithms generally irrelevant or weakly relevant to the scale of the scene, which is the inevitable trend as well.This thesis adopts Quadtree segmentation as technological school of LOD algorithm on the basis of studying various kinds of existing algorithms, and the thesis has greatly improved it and has overcome the insufficiency of traditional LOD algorithms and can make it meet the need of random terrain size without being restricted by the traditional Quadtree algorithms which must meet the size of (2~n+1)*(2~n+1), and at the same time introduces eigenvalue technology and greatly increases simplification efficiency. Additionally, the thesis has also introduced the technology of block storage and dynamic scheduling of the terrain data, and adopts multithreading and predictive pre-reading data strategy so as to make the systematic efficiency basically free of the size of terrain range and meet the infinitely great need of terrain range. Finally, on the basis of the technology, a practical system of roaming is realized.