Glander, Tassilo; Trapp, Matthias; Döllner, Jürgen
This paper discusses concepts for the automatic generalization of virtual 3D landscape models. As complexity, heterogeneity, and diversity of geodata that constitute landscape models are constantly growing, the need for landscape models that generalize their contents to a consistent, coherent level-of-abstraction and information density becomes an essential requirement for applications such as in conceptual landscape design, simulation and analysis, and mobile mapping. We discuss concepts of generalization and working principles as well as the concept of level-of-abstraction. We furthermore present three exemplary automated techniques for generalizing 3D landscape models, including a geometric generalization technique that generates discrete iso-surfaces of 3D terrain models in real-time, a geometric generalization technique for site and building models, and a real-time generalization lens technique.