Geometry-aware fine-grained decomposition of Chinese-style buildings from aerial photogrammetric 3D models

Fine-grained decomposition of Chinese-style buildings from aerial photogrammetric 3D models remains challenging because of elongated, interconnected, and geometrically complex components. This study proposes a geometry-aware framework for the fine-grained decomposition of Chinese-style building components from aerial photogrammetric 3D models. The framework integrates a geometry-adaptive segmentation network for extracting elongated architectural components from multi-view projective images, a geometry-preserving semantic transfer strategy for transferring high-resolution 2D semantics onto reconstructed 3D models, and a geometry-constrained decomposition method for separating interconnected roof components. In addition, a benchmark dataset with annotations of elongated roof components is constructed for evaluation. Experiments demonstrate that the proposed framework improves the extraction of elongated architectural components in both multi-view images and reconstructed 3D models, achieving more structurally complete, topologically consistent, and geometrically accurate decomposition results. The proposed framework and dataset provide a foundation for future research on digital documentation and structural analysis of heritage buildings with similar characteristics.