Polygonal representations have proven to be an efficient means of representing terrain environments at current levels of fidelity. Regular Triangulated Networks (RTNs) are a computationally-efficient means of capturing all posts in an elevation raster, while Triangulated Irregular Networks (TINs) can provide a more natural terrain appearance with fewer polygons. Both of these representations have done well at modeling terrain from sensor data with 1-meter or larger ground separation distance, but what happens when such data are available at decimeter, centimeter, or even sub-centimeter resolution? Is there a "cross-over" point where non-polygonal representations can capture 3D gridded earth measurements with less storage, and/or in a manner conducive to faster queries or rendering? This paper will compare the effectiveness of polygons, point-clouds, and voxels at representing ultra-high resolution terrain environments. We will compare the storage footprint of a high-resolution terrain data set in point-cloud, polygonal, voxel-grid, sparse voxel and octree voxel forms. modeling and simulation.
展开▼
