Multiresolution 3-dimensional terrain modelling

Abstract

With the introduction of high volumes of terrain-related data, including elevation data, cultural feature maps and satellite imagery, automatic terrain model generation has become a very popular topic for the modeling and simulation community, especially in mission planning, rehearsal and debriefing applications. Terrain modeling involves excessive amounts of data that raise performance issues for computer graphics hardware. The tradeoff between high quality, high fidelity geometry and high render frame rates can be achieved by the use of multiresolution modeling techniques. These techniques involve creating multiple representations of the same region or object with varying levels of detail (LOD), and selecting, or on-the-fly generating, the appropriate version of each region or object for each frame being rendered. The paper describes the software environment we have developed to generate automatically high-quality terrain models with varying levels of detail. In this environment, we have implemented many geometrical algorithms, such as constrained Delaunay triangulation and view-independent progressive mesh generation. Our current work focuses on using contour maps to develop a new algorithm to obtain multiresolution 3D terrain models.