Mathematical morphology-based generalization of complex 3D building models incorporating semantic relationships

A complex 3D building model contains a detailed description of both its appearance and internal structure with authentic architectural components. Because of its high complexity and huge data volumes, using a less detailed representation for the distant visual application of such a model is preferable. However, most mesh simplification algorithms cannot preserve manmade features of such models, and the existing 3D generalization algorithms are mainly proposed for regular-shaped buildings. More importantly, neither method can consistently express geometry, topological relations, and semantics in multiple discrete Levels of Details (LoDs). This paper presents a novel mathematical morphology-based algorithm that generalizes the complex 3D building model in a unified manner using the following steps: (1) semantic relationships between components, which reflect structural connectivity in the building at a certain LoD, are defined and extracted; (2) semantically connected components are merged and trivial geometric features of the components are eliminated simultaneously, with semantics associated with components then updated according to the merging; and (3) post-process is carried out to further reduce the redundancy of facets. The semantic relationships extracted ensure the proper generalization of topological relations and semantics of building components, and mathematical morphological operations implemented in the algorithm are capable of handling closed two-manifold components of various shapes. Experiments on both complex 3D building models in the classical Chinese style and prismatic 3D city models prove the effectiveness of the proposed method.