3D building mesh simplification based on structure perception

Simplifying three-dimensional building models is a key technology in urban digitalization. Existing general simplification methods struggle to preserve the main structural features of buildings, and specialized building simplification methods have limited applicability. Therefore, this paper proposes a structure-aware three-dimensional building mesh simplification method. This method achieves structure-aware plane segmentation by calculating the planarity of a k-ring neighborhood. A planar topology graph is constructed to express the topological relationships between planes. Based on multiplane constraint identification and optimized corner point locations, constrained Delaunay triangulation is used to achieve structure-preserving mesh reconstruction. Experimental results show that this method maintains a small Hausdorff distance while maintaining an average compression rate exceeding 80%. This method effectively preserves the geometric structural features of buildings, providing an effective technical approach for the efficient visualization of three-dimensional building models.