Abstract
This article presents a cartographic generalization approach for three-dimensional (3D) building models with different roof structures. Five different levels of detail (LODs) are defined in City Geography Markup Language (CityGML). CityGML does not indicate methods for the automatic derivation of the different LODs and approaches that can generate coarser LODs from detailed automatically should be developed. This article presents an approach for deriving LOD1 and coarser LOD2 buildings from detailed LOD2 models. In generalization of LOD2 buildings, the buildings are divided into three main types: basic, combined, and complex. F-type, G-type and P-type described by plane polygons, structured lines, and structured points are introduced to handle the common roof types. Seven combined building types are formed by the three basic types. As many buildings are modeled without semantic information in LOD2, roof, wall and ground faces are identified from building surfaces firstly and then the buildings are then recognized automatically by a progressive method according to their different roof structures. Different operators are designed to simplify different roof structures. In the simplification of combined roofs, new algorithms considering the spatial relationship of building parts are developed. Furthermore, aggregation algorithms are developed to address basic buildings distributed by certain patterns. Finally, buildings models are reconstructed based on the roofs and ground.
Acknowledgements
The authors would like to thank the editors and the anonymous reviewers for their valuable comments and suggestions, which have helped to improve the context and presentation of the paper.