Abstract. With the rapid development of digital technology and urbanization, indoor 3D reconstruction plays a crucial role in the construction of smart cities. This paper proposes a multi-level point cloud modeling method for complex indoor scenes based on spatial analysis and semantic enhancement. Firstly, room-level semantic segmentation is achieved by combining legal vectors and density features; Then, the α-Shape algorithm is adopted to extract the room contour, and the corner points are identified based on the local geometric features to refine the room boundary and wall structure; Next, the wall point cloud is classified and its attributes are determined through buffer analysis and surface-plane fitting to distinguish between planar walls and curved walls. Finally, the wall line structure is optimized by using the geometric regularization strategy, and each room is integrated into a complete indoor structure model with the help of spatial topological relations.The experimental results show that this method significantly improves the modeling accuracy and robustness in non-Manhattan structural environments, and can accurately reconstruct the geometric and semantic information of complex indoor Spaces. This provides a solid data foundation for indoor 3D reconstruction and intelligent building applications.
Liu et al. (Wed,) studied this question.