| 3D digital earth, associating with multi-subjects, is a complex systematic engineering. It has been made some progresses, but it is difficult to attain the effects of real-time display and interactive manipulation. The technical challenges are to the construction and management of multi-resolution large volume spatial data models, the efficiency of spatial indexing, limited computer network transmission, construction and visualization of complex 3D digital city models, seamless integration of multi-source data, real-time walkthrough and interaction of large volume scene, spatial analysis and queries.Aimed at the challenges of developing digital earth, the dissertation discusses some key technologies, such as the methods of dynamic visualization and walkthrough of large volume DEM, vector data, imagery and 3D models on the Internet. The dissertation emphasizes on large data transmission on the network, visualization, spatial analysis and system architecture. Based on the theories and techniques introduced in this dissertation, the group has designed and developed a 3D digital earth platform. The platform will finally implement multi-source data visualization and seamless integration, 3D digital city reconstruction, meteorology monitoring, war field simulation, navigation, querying etc. The platform also supports running on ordinary computers. It was praised as an excellent 3D image information platform by ministry of science and technology, China, and it has been used in some departments, such as military sections.Based on the existing achievements, the author organizes the whole dissertation in which the 3D digital earth system we developed is a main thread, and the theories are elements. The dissertation has mainly the following innovations:1. The dissertation puts forward to use the lifting scheme to construct a second-degree average interpolating wavelets and then use them to create multi-resolution models. The algorithm need fewer computations and can maintain the created models being high precise. It also overcomes the disadvantages that the pyramid structure can't be reconstructed precisely.2. In order to have access to the very large volume spatial data as fast as possible, a proper spatial indexing is put forward. In the dissertation, the ellipsoid quadtrees described by Ottoson are improved and introduced for indexing the large volume spatial datasets.3. This dissertation firstly presents an optimal predictor to identify and removes the spatial redundancy among the elevation data sets and then proposes an effective scheme based on arithmetic coding to encode the residual elevation data. Results from numerical experiments on real datasets are given to demonstrate that the proposed novel lossless data compression scheme provides a proper way for elevation data compression.4. The predicting model of JPEG_LS is improved, which makes the predicting precise further enhanced. Lossless image compression by JPEG_LS is implemented.5. An approach is described to create view-dependent, multi-resolution terrain models from large raw height fields based on M-band wavelets and bintree triangles in this paper. The algorithm gives a way of high efficient data compression and regular grid datasets simplification depending on surface roughness, viewpoints, view directions and system load conditions. The experimental results indicate that the process has much smaller memory requirements, displayed terrain scene at high interactive frame rates without compromising visual quality. The algorithm also overcomes the 'cracks' caused by the neighboring different resolution terrain quadrangles.In addition, the author has studied on the following questions:(1) WGS84 geodetic datum and GRS80 reference ellipsoid are used to construct the earth model. The algorithm of the ellipsoid surface culling inside the view frustum is discussed.(2) Research on the method of seamlessly overlaying linear vector and planar vector data on digital elevation models and imagery. Aimed at the geometric properties of l...
|
PDF Full Text Request