Wavelet-based Large-scale Topographic Data Organization With Continuous Multi-resolution Performance

As is universally acknowledged, 3D Terrain Roaming System plays an essential role in such fields as Virtual Reality and Geography Information System, etc.. Moreover, the technology of Terrain Visualization has also been one of the central issues of terrain-correlated field these years.However, the uninterrupted enhancement of scale and requirement in various application systems of terrain(especially large-scale ones) and the exceedingly massive terrain data have gone beyond the ordinary graph's capability of real-time rendering and memory management. Consequently, huge amounts of data make up complex models of terrain surface, while the computer's figure hardware is of limited rendering ability. Thus solving the conflict between the complex models and the limited rendering ability has become the core issue of terrain visualization. Furthermore, on large-scale terrain application, except for the simplification of rendering, new demands have been put forward on the aspects of data storage and data loading.This dissertation was based on the analysis and research of contemporary terrain rendering technology and wavelet multiresolutional analysis. Taking the characteristics of large-scale terrain application into account, this paper carried out discussions centering on the two aspects of mutiresolutional rendering and terrain data memorization & dispatch:1 , This paper studied the theory of wavelet multiresolutional analysis , the data structure of Digital Elevation Model (DEM), and illustrated that DEM may be seen as the terrain's 2-dimentsion signal. It also proposed a model of processing DEM data multiresolutionally with the methods of wavelet.2, View-dependently, distinctive multiresolution were applied to different terrain blocks within a certain perspective. According to the view distance, this paper proposes a method that terrain meshes with various details were dynamically reconstructed in a continuous way by using wavelet. Thus smooth shift between multiresolutional terrains with the change of view-point distance was realized. Additionally, the "crack" problem between terrain blocks of different multiresolution were also successfully solved.3, In view of the significant amount of data that a large-scale terrain has, and the characteristics of the after-wavelet-transformation terrain data, this dissertation explored an out-of-core data loading & storage method. This method doesn't need any extra storage space, and applied a multithreading dispatching algorithm of pre-loading and real-timeloading, which can raise rendering efficiency greatly.4*. Finally, the above-mentioned method were verified through an experiment. As the result demonstrated, continuous multiresolutional effect can be achieved and the requirement of real-time rendering can be met.