Large Terrain with High Resolution Aerial Photographs

B.-Y. Li, H.-S. Liao, C.H. Chang, and S.-L. Chu (Taiwan)


Visualization, Large Terrain, VR, LOD, GIS.


Interactive navigation of large terrain data with high resolution aerial photographic image has been a challeng ing problem. In this paper, we propose the level-of detail(LOD) modeling with quad-tree[1] to speed up the terrain mesh rendering and using tiled texture[2] to dis play the detail of the aerial photographic image in real time rendering. The terrain data is divided into blocks to be stored in the nodes of the quad-tree and the quad-tree will be dynamically traversed to decide which nodes to be shown. Between two adjacent blocks, a dike structure is proposed to provide a smooth blending between two blocks of meshes with different levels of detail, and therefore re move the possible artifacts and cracks that usually occur when adjacent blocks have different levels of LOD data. The high-resolution aerial photographic images are divided into tiles, and will be pre-processed into several levels of detail similar to the geometric LOD described above. Many other mechanisms to speed up the perfor mance are used, such as view culling. Our proposed meth ods will be useful in building a terrain data rendering sys tem for flight simulation and geographic information sys tem (GIS) applications which have relatively large size of polygons with gigabytes of aerial photographic images.

Important Links:

Go Back