Christopher Weber

• Today, polygonal models occur everywhere in graphical applications, since they are easy to render and to compute and a very huge set of tools are existing for generation and manipulation of polygonal data. But modern scanning devices that allow a high quality and large scale acquisition of complex real world models often deliver a large set of points as resulting data structure of the scanned surface. A direct triangulation of those point clouds does not always result in good models. They often contain problems like holes, self-intersections and non manifold structures. Also one often looses important surface structures like sharp corners and edges during a usual surface reconstruction. So it is suitable to stay a little longer in the point based world to analyze the point cloud data with respect to such features and apply a surface reconstruction method afterwards that is known to construct continuous and smooth surfaces and extend it to reconstruct sharp features.