Boundary regularization of point clouds facades based on terrestrial laser scanning data

Traditional methods for boundary regularization mainly take airborne laser scanning (ALS) data as input and process the point clouds of building roofs with simple planar shapes, while terrestrial laser scanning (TLS) could acquire more complete and denser point clouds of building facades with more complex shapes. In this paper a new boundary regularization approach based on TLS data is proposed. It can deal with several types of point clouds for buildings facades, including walls, doors and windows. Firstly, raw point clouds are segmented into planar facades by using an efficient RANSAC algorithm. Secondly, the boundary points are extracted by using a 2D alpha-shapes algorithm. Based on the boundary points, a fitting regular polygon is generated by using the boundary regularization approach proposed in the paper. The experimental results show that the proposed approach is self-adaptive to point clouds with different density. Meanwhile, it is demonstrated that the approach is effective and efficient, and could provide reliable and satisfactory regularization results.