Software for Generating Geometrically and Topologically Accurate Urban Terrain Models Using Implicit Methods

Currently, the terrain models in most widespread use consist of surfaces represented explicitly as planar triangular facets connected in a continuous manner with each other (triangular irregular networks or TINs). Generally, these models are "terrain skins," that is, representations of height as a univalent function of two independent lateral variables (such as latitude and longitude). Objects with multivalent height and non-genus-0 topology are often added onto these terrain skins to make the terrain model approximate reality closer. However, for urban terrain, the number of such objects and the corresponding computational burden are huge. Raw data that corresponds to objects with multivalent height and non-genus-0 topology (bridges, overhangs, underpasses, buildings, subways, etc.) is becoming widely available. During this effort, a fast and accurate software package for generating implicit representations of multivalent, non-genus-0 urban-terrain from point-cloud data will be developed.