520 |
|
|
|a The massive volume of data involved in the development of a real-time terrain visualization and navigation system is inevitable. Based upon the current offthe- shelf hardware capability, it is impossible to process the amount of data using the conventional approach. This is due to the fact that the amount of data to be processed exceeds the capacity that can be loaded into main memory. This problem is further compounded by other hardware constraints such as memory bus speed and data transfer bandwidth from the main memory to the graphics card. Consequently, this limitation has affected the total system performance. The triangle-based level of detail (LOD) technique has been developed in order to reduce the drawback but it still suffers from the main memory constraint and slow data loading. The purpose of this research is to design, develop and implement an algorithm for enhancing the rendering efficiency of triangle-based LOD technique. A prototype system has been developed using digital elevation data for testing purposes. The system was evaluated based on the following criteria, namely: data size, processing time for data partitioning, memory usage, loading time, frame rate, triangle count and geometric throughput. According to the results obtained during the pre-processing, the partitioning of digital elevation data into tiles has successfully reduced the data size although it required a longer processing time. During the run-time processing, the integration of dynamic tile loading scheme, view frustum culling and enhanced triangle-based LOD technique has produced encouraging results with significant overall improvement as compared to the techniques that have been developed earlier. The proposed algorithm in this thesis is very practical in developing interactive graphics applications, which consider real-time rendering as one of the important elements.
|