Data Block Partition and Database Based Large Scale Terrain 3D Visualization

This paper presents a database-based optimized algorithm for terrain data organization, storage, scheduling and multi-resolution rendering; besides, some technical means in data dynamical scheduling are illustrated, such as data encoding and indexing, view frustum culling, cache management and multi-threading. Due to the drawbacks of ROAM algorithm when dealing with large scale terrain data, ROAM is only applied to individual block to build multi-resolution LOD models. The evaluation function, cracks elimination, triangle strips and some other key questions are discussed as well. Experimental tests are conducted on the efficiency of terrain visualization in a LAN environment, results indicate a better performance on terrain rendering and fast roaming, this optimized algorithm can meet the requirements of large-scale terrain visualization.