This paper describes an approach for the typification of buildings using a mesh-simplification technique. The approach is adapted from the area of computer graphics and was originally developed for surface reconstruction and mesh simplification. The main goal was to develop an algorithm which creates fast and reproducible results. The typification procedure is modelled as a two-stage process, with the steps 'positioning' and 'representation'. While the positioning step determines the number and the position of the building objects based on Delaunay triangulation, the representation step is used to calculate the size and orientation for the replacement buildings. The results presented show the important influence of weights during positioning steps to control the object distribution. The proposed parameters are the number of objects as well as several object characteristics such as size, shape, orientation, and semantic. The approach has to be extended, if building alignments are also to be preserved. Further applications are imaginable, for instance the icon placement on dynamic maps.