Research On Key Technologies For 3D Visualization Of Large-scale GIS Data

With the rapid development of computer and information technologies, 3D visualization technology for geographic information is applied in more and more fields. 2D Geographic Information System (GIS) is essentially based on abstract graphic symbols, which is lack of such aspects as sense of reality and interaction. Therefore, it is difficult for 2D GIS to reflect, analyze or even display the relevant information precisely when dealing with 3D problems such as electromagnetism, meteorology and geology, while 3D visualization of the geographic information usually makes these problems clear and intuitive.So far, there is no standard model of 3D spatial data or unified framework for visualization of it. Comparatively, 2D GIS has been widely researched for decades, and it contains abundant geographic information with very low price. This thesis focuses on realizing fast construction and efficient rendering of 3D virtual environments based on GIS data. In detail, the main work and contributions are described as follows:A visualization framework for large-scale terrain based on GIS terrain data is proposed after thorough research of 3D visualization of large-scale terrain scene. The proposed framework can meet the requirements of both 3D visualization of large-scale GIS terrain data and the integration of vector data.Based on algorithms for hierachical GIS vector data layers generation and its simplification described in detail, an approach for constructing, organizing and scheduling large-scale GIS vector data is proposed. The proposed approach aims at resolving the contradiction between the large size of vector data and the limitation of computer's real-time processing ability.A method for 3D visualization of GIS vector data is proposed. 3D terrain features based on vector information are firstly classified by their characteristics, then either feature adjusting or terrain adjusting method is chosen for matching of features and terrain. The algorithms also deal with some special cases such as low resolution of terrain dataset and multi-vector restriction applied to the same terrain area.Visualization of raster and vector data is quite different because of the difference of their inherent structure. Based on the above practice of our 3D visualization algorithms, a framework for visualization of 3D virtual environments based on GIS data is proposed. The proposed framework gives reference for the design of 3D visualization system based on GIS data.A 3D visualization of large-scale GIS data system named GeoSprite is designed and implemented, which gives a sound support to the above framework and algorithms. The GeoSprite system could also be applied in many other applications as a fundamental platform.Summarily, this thesis studies key technologies on 3D visualization of large-scale GIS data, implements fast construction and real-time rendering of 3D virtual environments based on large-scale GIS data, which build solid ground for other relevant applications.