Research On Modeling And Visualization Technology For Deep Space Surveying And Mapping Spatio-temporal Data

In combination with the related research work of the 863 Program and the pre-research of the 11th five-year plan, in the thesis, the theory and algorithms of deep space spatial-temporal data modeling and visualization are studied. The involved key techniques include the spatial-temporal reference, the spatial-temporal data model, real-time visualization of planet terrain data, volume rendering of space environment elements. A prototype system of moon-earth space environment visualization is realized.The main work stated in this thesis focuses on the following aspects.1. The definition, task commandment and key techniques of deep space surveying are presented and analyzed. The spatial-temporal references of deep space surveying are summarized and classified. According to the characteristic of deep space surveying and mapping, an idea for the different spatial-temporal reference transformation is presented and some important reference transformation formulas are induced.2. Based on the analysis of the composition and sort of deep space environment's spatial-temporal data, the sort scheme for the space objects is proposed. The basic spatial-temporal evolve process is presented for space object and the Deep Space Moving Objects Spatial-Temporal model—DSMOST is designed and realized. The data model can describe the space object's geometry structure, exterior movement and interior behavior. Aiming at the requirement of space environment visualization, a spatial-temporal partition model OSAT is proposed based on the octree space division and adaptive time division.3. Six typical terrain LOD(Level-of-Detail) algorithms are summarized and the CLOD (Continues Level-of-Detail) algorithm is studied and realized. Great improvement is made on the Geometry Clipmaps algorithm. And a new asymmetry Clipmaps algorithm is put forward. Experiment results demonstrate that the algorithm can improve the terrain rendering effect with increasing little data.4. To realize the real-time out of core rendering for the global terrain, an LOD algorithm based on the variable quadtree is proposed. The algorithm integrates the excellency of CLOD,Chunk-LOD and Geometrical Mipmapping algorithms and increases the terrain details by extending the Quadtree.5. Based on original lunar terrain data, the lunar surface details is increased by fractal technique, lunar crater and rock simulation. Relax quadtree was applied to improve search and cull efficiency of lunar rock.6. A two-dimensional transfer function based on the volume data histogram and the scalar data second derivative is proposed. The space environment volume data can be transferred into color and opacity in the two-dimensional transfer function. The global space environment data is rendered in 3D texture volume visualization and ray-casting volume visualization respectively. An acceleration rendering scheme based on Octree texture is proposed.7. The atmosphere lighting model is studied and the integral over the ray is computed in the Graphics Processing Unit by CG program. The sampling spheres, texture lookup table, back face culling and view range culling are used to improve the rendering performance.8. On the basis of the above theory and algorithms, a moon-earth space environment visualization prototype system is constructed and moon-earth surface multi-resolution rendering, space environment volume data rendering and planet, satellite's movement simulation are also realized.