3D Building Generalisation

The rapid development of multisensor and multimedia technologies has made it possible to construct and visualize detailed three-dimensional (3D) city models. 3D city models are typically rendered as central perspectives with rich depth cues and a self-explaining character. They offer an intuitive organization of spatial objects that replicates or reflects the real world, thus utilizing the viewer's natural perception and memory of space as well as spatial relationships. This chapter provides arguments related to the need for automatic 3D generalization. It provides a basis from which new techniques can be developed as well as highlights the need to extend 2D operations such as aggregation, typification, and landmark exaggeration to cope with 3D generalization problems. Although these application fields share the common demand for 3D information, their special requirements considerably differ with regard to precision, actuality, spatial coverage, and interoperability. Most of the available methods have focused on the geometric simplification of a limited set of sample buildings. A key goal of generalizing 3D buildings is not merely to reproduce a physical form, but to enhance the process of spatial cognition by making characteristic parts more salient.