Database support for multi-resolution terrain models

Xu, Kai (2004). Database support for multi-resolution terrain models PhD Thesis, School of Information Technology and Electrical Engineering, The University of Queensland.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE17869.pdf Full text application/pdf 9.61MB 0
Author Xu, Kai
Thesis Title Database support for multi-resolution terrain models
School, Centre or Institute School of Information Technology and Electrical Engineering
Institution The University of Queensland
Publication date 2004
Thesis type PhD Thesis
Supervisor Prof. Xiaofang Zhou
Assoc. Prof. Xuemin Lin
Total pages 214
Collection year 2004
Language eng
Subjects L
280103 Information Storage, Retrieval and Management
700100 Computer Software and Services
Formatted abstract

Terrain data of very large scale is widely used in many applications, such as virtual reality, environmental analysis, and GIS. The large size of terrain data makes it very demanding on memory and processing power during visualization and spatial analysis. Multi-resolution techniques have been introduced to address this problem. A Multi-Resolution Terrain Model (MRTM) can provide a large number of simplified approximations {meshes) at various resolution levels together with the original model. For many applications, MRTM can reduce the required data amount by replacing the original terrain model with a mesh at a lower resolution. Such a mesh contains much less data but can be accurate enough according to application requirements.      

The multi-resolution techniques are applied for both terrain visualization and spatial analysis. However, most existing approaches are based on main memory, i.e., they assume that the entire MRTM can fit in the memory. The fast growing rate of terrain data makes this assumption increasingly impractical. These methods suffer from excessive overhead of data retrieval and do not scale with the data size. The data exchange between the main memory and secondary storage becomes the bottleneck for many visualization and spatial analysis applications using MRTM.      

In this thesis, new methods are proposed to support secondary-storage MRTM applications. To support various functionalities in visualization and spatial analysis, these methods focus on Selective Refinement (SR), which is a fundamental operation found in most of these applications. After a thorough study, a key problem in secondary-storage SR is identified and desirable properties of possible solution are provided. Based on this discussion, the first method, the LOD-quadtree, introduces a new dimension into the data space and provides a novel indexing structure designed according to the data distribution of MRTM. This method improves the I/O performance of SR considerably by reducing the number of retrievals during SR. As an adaptive method, LOD-quadtree can be applied to various MRTMs proposed in the literature. The second method. Direct Mesh (DM), further improves the performance of SR for a popular type of MRTM called Progressive Meshes (PM). The enhancement of I/O performance is achieved by improving the data structure of the PM. After a careful study, selected topology information is encoded at each node in the PM with reasonable storage overhead. As a result, the amount of data required for a SR is significantly reduced. The cost model of DM is provided, together with an algorithm that further reduces the data retrieval cost for meshes of varying resolution.  

Keyword Spatial analysis statistics -- Data processing
Numerical grid generation (Numerical analysis)

Document type: Thesis
Collection: UQ Theses (RHD) - UQ staff and students only
Citation counts: Google Scholar Search Google Scholar
Created: Fri, 24 Aug 2007, 18:26:28 EST