QLOD: An Adaptive Multiresolution Structure for Terrain Visualization

This work introduces a new hierarchical structure to interactively visualize large terrains. Mainly two different approaches have been used in terrain visualization: algorithms based on regular structures (providing a natural frustum culling) and algorithms that use irreg-ular meshes highly adapted to terrain features such as ridges and crests. In this work, we present a new structure which combines the advantages of a quadtree with an adaptive irregular triangulation. With the quadtree it is possible to achieve very fast frustum culling with a global hierarchical level of detail (LOD) decision, while the irregular mesh, well adapted to the terrain, provides a local LOD decision. QLOD enables efcient mesh extraction, representing the terrain adapted to the point of view. Our structure also guarantees a continuous, smooth transition between the levels of the quadtree, seamlessly inserted into an irregular mesh. We also discuss an im-plementation of these ideas and a series of tests using large terrain models, which lead to several conclusions