Progressing Towards Interoperability: Using the Lifting Scheme for Multi-resolution Terrain Databases

摘要

Multi-resolution terrain representations were first proposed to accommodate the needs for storing and processing the same terrain models at different resolutions. With one set of the source data, many resolutions can be produced and then rendered intelligently to reduce the graphical load. To accomplish this purpose, the terrain model is pre-generated at different resolutions and stored in the database. Typically, when being rendered, the visualization application loads into memory a set of terrain sections at the appropriate resolution for a given viewpoint. The terrain model should allow for processing and rendering at any arbitrary resolution at runtime, depending on the requirements of the interoperating applications. In this research, we develop a common multi-resolution terrain model that enables dynamic rendering and processing with the eventual goal of facilitating interoperability among Computer Generated Forces (CGF) for constructive and virtual systems and visual simulation systems. A prototype of our method demonstrates significant improvements in storage space and construction time while still maintaining good accuracy when compared to some other widely used approaches. Experiments with the Grand Canyon database shows consistent results in the reduction of storage space, using as little as 1/5 of the disk space when compared to Hoppe's Progressive Meshes, while the RMS error of the worst-case scenario is still 1/2 of the best case in the ITIN work of Polis and McKeown. The prototype also demonstrates the dynamic processing and rendering capability of this compact terrain model that shows the potential for improved interoperability over a wide range of networked interacting applications.