Optimized 3D Building Mapping and Reconstruction via Cross-View Collaboration in Densely Built-Up Areas

High-precision urban three-dimensional (3D) point clouds can be effectively generated using airborne oblique photogrammetry and lidar scanning. However, due to occlusions from dense building layouts and vegetation, existing airborne acquisition methods often struggle to capture complete building geometries. This incompleteness poses serious challenges for applications such as urban planning, smart city management, and autonomous navigation, which require structurally complete and accurate 3D data. To address this limitation, this paper proposes a novel cross-view collaborative surveying framework that integrates aerial and ground-based data collection for precise and efficient 3D reconstruction of urban buildings. The framework begins by performing automated building completeness detection on aerial point clouds using a multi-layer slice projection algorithm, which enables accurate identification of missing regions in both point-wise and surface-wise forms. These detected deficiencies are then used to guide the generation of optimized ground-based supplementary acquisition routes, incorporating a global-local planning mechanism and a multi-objective technologies for autonomous robot exploration (TARE) strategy to enable autonomous and adaptive data collection. Comprehensive experiments were conducted in both simulated and real-world urban environments. Evaluation metrics focused on point cloud completeness and 3D reconstruction accuracy. The results demonstrate that the proposed method significantly enhances the completeness of building point clouds, achieving an average detection accuracy above 90%, while also reducing reconstruction error by up to 65% in complex urban scenarios. The proposed method provides a valuable tool for urban mapping professionals, autonomous systems, and digital city infrastructure developers who depend on high-quality 3D building models.