Unstructured grid generation over buildings intersecting terrain data using a feature-preserving volumetric technique
Juan Raul Cebral, Fernando E. Camelli, Rainald Loehner · 2002
In this paper a new methodology to generate unstructured grids over intersecting objects is presented. It is based on the extraction of the zero-level iso-surface from a signed distance map defined on an adaptive background grid. The background grid points are first created following local object voxelizations and then connected using a Delaunay triangulation algorithm. A point enrichment strategy is then used to introduce new points into the background grid along detected surface features, i.e. ridges. The background grid is then adaptively refined close to the object surfaces and the distance to the surface function is computed. The final merged surface is obtained by extracting the zero-distance iso-surface. This watertight triangulation is then used as a support surface for unstructured grid generation from discrete data. These algorithms allow realistic modeling of a large number of complex engineering geometries using overlapping components defined discretely, i.e. via surface triangulations. This capability is very useful for grid generation starting from data originated in measurements or images.