Visualization For Multi-Resolution Restricted Quandtree Modeling Of Large Scale Terrain

Real-time visualization of the large scale terrain occupies an important position in real-time simulation and GIS. With the development of remote sensing technology and satellite technology, it is not impossible to get high-resolution digital elevation data and geometric image texture data. So, in order to achieve realism and real-time roaming of the large-scale terrain,we must build a reasonable system of data organization and scheduling, and make a multi-resolution model. This is also the research focus of international scholars in terrain simulation. Aims of this article are to build large-scale multi-resolution terrain model and to achieve the result of realistic real-time roaming. The topic on the large-scale terrain visualization has important theoretical and practical significance.Firstly, based on reading and analyzing a lot of classic terrain modeling theories and algorithms, we designe hierarchical multi-resolution quadtree block structure to organize the terrain data, using hierarchical multi-resolution terrain block structure and combining the restricted quadtree theory proposed by Pajarola.Secondly,in order to make the multi-resolution restricted quadtree structure be used in wider range of applications, we must consider the affect of the view-point speed and the terrain importance to roaming results. We introduce the roaming speed and the terrain importance factors to reconstruct the terrain node evaluation function. The new function ensures that the more attention of the terrain the lowest distortion in the real-time terrain rendering. We introduce GPU-based occlusion culling algorithms to minimize the idle time between CPU and GPU. In order to ensure the terrain grid get over smoothly and seamlessly, we use merging and segmentation method which is based on view-point distance to eliminate cracks.Thirdly, in order to improve the efficiency of the terrain rendering, the quantitative compression method of positon informations about vertex data was designed based on the terrain model. Besides, we also presented a new triangle strip method to organize vertex datathat. With these methods, the redundant data need to be transferred were reduced.Finally, to verify the validity of multi-resolution terrain model, we designed and implemented a large-scale terrain visualization system. Experiments showed that the system can achieve higher frame rates and better rendering results, and achieve real-time visualization roaming of the large-scale terrain.