Abstrakt
Many representations of topographic surface providing graphical, three-dimensional, multi-resolution model of entire planet have been developed. However, current solutions to such a representation of topographic surface suffer from the lack of geometric flexibility and accuracy on boundaries with models of other geographic features.
Therefore, this work focuses on a more functional representation of a multi-resolution topographic surface. For this sake we introduce a simplification algorithm, which is designed to build the multiple LOD database of features.
The method utilizes the Global Indexing Grid (GIG) as a paging mechanism. For any position of the observer within the 3D virtual environment, the indexing structure determines the position-dependent LOD of currently visible features and underlying terrain.
The simplification algorithm guarantees for any observer position the preservation of topological relations between simplified geometries of features in the position-dependent LOD visualization of the Earth surface. Based on the precomputed classification of the elevation points and multiple LOD database of features, the surface is reconstructed using constrained Delaunay triangulation at run-time.