Real-Time Terrain Rendering And Optimization For Large-Scale Complex Scene

Since 1990 the technology of virtual reality has always been one of hot topics in computer graphics field. 3D real sense terrain is an indispensable factor to the virtual natural environment, and it's also an important part of the virtual simulation research field. But because of the limitation of hardware technique and data transport technique, real-time display complex terrain scene and real-time process large-scale terrain data become technical barrier in terrain generation technology. Therefore, how to describe a large-scale complex terrain scene with reality and realize drawing of a real-time terrain scene is a domestic and foreign research hotspot.Real-time terrain rendering and optimization are mainly discussed and studied in this paper. The technologies of visibility culling and level of detail are applied. Combined with real-time optimal adaptive meshes algorithm, fractal algorithm, the quadtree recursive segmentation algorithm, GeoMipMap algorithm, some effective simplified strategies were proposed. Application fields of simplify technology for terrain can be divided into two categories.In the case of using the real data to establish terrain model and dynamic roaming, Real-time optimal adaptive meshes algorithm was improved to accelerate drawing frame rate and optimize terrain rendering. In this simplified solution, visibility culling technology was further simplified and screen error criteria was used by combining nested bounding sphere method with screen error method.In the case of simulating a real sense virtual terrain scene where real data were not used, Use fractal algorithm to create complex topography image with small amounts of data, and analyze the parameter and optimize rendering to greatly enhance the fidelity of the image.For fractal algorithm has limitation in level of detail aspect, it was combined separately with dynamic quadtree algorithm and dynamic GeoMipMap algorithm which have level of detail. The two methods calculated simply and the image with high fidelity. The terrain scene not only has randomness and authenticity, but also has dynamic and real-time character.All the strategies suggested in this dissertation are explained in detail in theories. Correctness and real-time character of the simplified algorithm are verified in simulation.