interior reconstruction using the 3d hough transform

Laser scanners are often used to create accurate 3D models of buildings for civil engineering purposes, but the process of manually vectorizing a 3D point cloud is time consuming and error-prone (Adan and Huber, 2011). Therefore, the need to characterize and quantify complex environments in an automatic fashion arises, posing challenges for data analysis. This paper presents a system for 3D modeling by detecting planes in 3D point clouds, based on which the scene is reconstructed at a high architectural level through removing automatically clutter and foreground data. The implemented software detects openings, such as windows and doors and completes the 3D model by inpainting.