Research On Real-Time Rendering Technology For Large-Scale Terrain Based On LOD

Real-time rendering technology for Large-scale terrain is becoming a hot topic in visualization,3DGIS,which attracts more and more researchers' attention at present.It has been widely used in computer graphic,virtual reality,terrain visualization and outdoor games etc.With the increasing of complexity of the scene,the amount of data to be rendered becomes larger.In order to improve the speed,LOD is used in the terrain rendering.This thesis compares and analyses the existing terrain rendering algorithm based on CPU and GPU,two new methods are proposed in the paper.Major research work can be summarized as follows:1.Key technologies in LOD Large-scale terrain real-time rendering are researched. Relative technologies are presented such as:LOD of the terrain,visibility culling,cracks eliminate,view-dependent refine,LOD error,technologies of texture etc.2.By researching the traditional continue LOD algorithm based on Quadtree,a new simplification algorithm is proposed.A new radar method which differs from the traditional six-surface method is used for view frustum culling and improves static nodes evaluation and boundary crash eliminating.This method simplifies the algorithm and improves real-time rendering speed.3.Texture blending,real-time water,scene of the fog are discussed in detail,those technologies are common used in outdoor scenes.Based on the new Quadtree algorithm, we design and realize a large-scale terrain real-time walkthrough system.4.A new algorithm is proposed that through researching the traditional algorithm based on GPU.This paper improves view dependent geometry clipmaps and normal blending,a regular nested grids which likes geometry clipmaps is used to speed up.In addition,an external texture memory support is provided to enable the rendering of terrains that exceed the size of texture memory.We also performed a comparative analysis with other terrain rendering techniques.The experimental results show that the proposed algorithm is feasible and efficient.