Abstract:The current building surface reconstruction from airborne LiDAR point clouds pays more attention to the relationships between roof primitives but ignores the possible existing points of walls and the structural relationships between the roofs and walls. In this paper, we propose a novel approach that reconstructs a building surface model from airborne LiDAR point clouds based on improved structural constraints. Unlike existing hypothesis-and-selected-based methods, which impose the same constraints on all relationships between planar primitives, we design a series of strong and soft structural constraints to promote the representation of geometric and structural details of both roofs and walls in the reconstructed models. To be precise, we generate a set of candidate faces with detected planar primitives of building and vertical planes inferred from the corresponding footprint to complete possibly missing walls. Then, the potential structural representations associated with building walls in the original point clouds are identified by conflict detection between the support planes both existing points of building walls and inferred from the footprint, and on this basis, improved structural constraints are constructed for the structures consisting of different type of support planes in the set of candidate faces. Finally, a compact and watertight model is extracted from the faces of the candidate set by energy minimization with proposed improved structural constraints. Experiments on the various airborne LiDAR point clouds and aerial point clouds datasets have demonstrated that the proposed method could obtain accurate and compact models with the representation of geometric and structural details between building roofs and walls compared with other state-of-the-art approaches.

Image-based building reconstruction with Manhattan-world assumption

A Flexible Method for 3D Reconstruction of Urban Building

Advanced-Technological UAVs-Based Enhanced Reconstruction of Edges for Building Models

City3D: Large-Scale Building Reconstruction from Airborne LiDAR Point Clouds

3D Building Reconstruction Based on Interpretation of Architectural Drawings

Multi-view Urban Scene Reconstruction in Non-Uniform Volume

Line-based single view 3D reconstruction in Manhattan world for augmented reality

Automatic Single View Building Reconstruction by Integrating Segmentation.

Hybrid Method of Connection Evaluation and Framework Optimization for Building Surface Reconstruction

Reconstruction of 3D Information of Buildings from Single-View Images Based on Shadow Information

Neural 3D Scene Reconstruction with the Manhattan-world Assumption

Real-time indoor scene reconstruction with Manhattan assumption

Fast Building Instance Proxy Reconstruction for Large Urban Scenes

Layer-Wise Floorplan Extraction for Automatic Urban Building Reconstruction

Point2Building: Reconstructing Buildings from Airborne LiDAR Point Clouds

Automatic Reconstruction of Building Façade Model from Photogrammetric Mesh Model

Line-Based 3D Building Abstraction and Polygonal Surface Reconstruction From Images

Semantic-guided 3D building reconstruction from triangle meshes

3D Residential Reconstruction from a Single Indoor Floor Plan.

Semantic decomposition and reconstruction of residential scenes from LiDAR data

3D reconstruction of building surface from airborne LiDAR point clouds based on improved structural constraints