Continuous Merging of Buildings Considering Functional Semantics and Global Optimization

Continuous generalization is a key technique for achieving continuous scale representation of map features, and the continuous merging of buildings is a crucial component of this process. Existing methods for continuous building merging predominantly focus on smooth geometric transitions, lacking explicit modeling of functional semantics and a search mechanism for globally optimal merging paths. To address this gap, this paper proposes a continuous building merging method that integrates functional semantic information with a global optimization strategy. The method first identifies building functions using POIs and AOIs data. Building aggregation patterns are then recognized based on Gestalt theory and serve as target constraints. Subsequently, a dynamic A* algorithm is employed to determine the globally optimal merging sequence, where the merging cost comprises distance, shape, functional semantics, and importance. Finally, buildings are merged sequentially through Delaunay triangulation filtering and orthogonalization algorithms. Experimental results on a Zhengzhou dataset show that the proposed method produces optimal merge sequences in 89.2% of cases and reduces the total merging cost by 12.6% over the best baseline, achieving continuous building merging with superior morphology and semantics.