Research Of Simplification For Large-scale3D Terrain

Currently, the three-dimensional visualization technology for large-scale terrain is concerned widespread, and has become the focus of the study. It is applied to3D geographic information system, virtual reality,3D scene games and many other areas. But it is difficult to handle massive terrain data efficiently under the limited hardware conditions, which makes improving terrain mesh simplification algorithms to be the most important research direction. On the premise of ensuring overall characteristics and precision requirements, simplification algorithms minimize the amount of data describing the terrain, improving the generation efficiency. Therefore, it is of great significance to research the large-scale terrain mesh simplification technology.Modeling a three-dimensional terrain includes two step, data processing step and generating and simplifying step. The huge amount of data is the fundamental reason to reduce the quality of terrain generation and roam. So effective data processing method is necessary to improve the effect of model simplification. And in the process of terrain partition, this paper presents a2-level dividing method. This paper presents a DEM data compression method using Morton coding in partition based on the surface roughness. Only the terrain blocks with a lower surface roughness is encoded and compressed, which could reduce the amount of reading DEM data and improve the compression speed. Besides, it also plays an important role to use effective generation method and simplifying rule. The quadtree data structure is discussed for regular mesh model simplification and view-dependent terrain generating rule is more focused on. The generating rule based on local terrain entropy is studied and improved to form a view-independent local entropy rule first. Then, combining the improved rule and Phong illumination model into our hybrid rule, we get a continuous and smooth terrain model. The presented rule improve the simplification rate, and simultaneously keep the local details of terrain.Finally, it shows that the both proposed algorithms are effective through experiments.