Automatic detection of doors and windows using point clouds in the context of as-built BIM

Abstract. The automation of 3D modelling from point clouds has emerged as a highly intriguing research area, offering significant potential value in the context of scan-to-BIM (Building Information Modelling) process. 3D modelling from point clouds is often performed manually, which is time-consuming and prone to errors. Most methodologies employed for the processing of building point clouds data rely on geometric information to extract objects, thereby ignoring radiometric information, which could provide a highly valuable addition to object detection. Emerging approaches integrate both geometric and radiometric data to enhance the robustness of the algorithms. The objective of this study is to develop an automatic method for the extraction and 3D modelling of building openings (doors and windows). The process makes use of both geometric and radiometric information, applied to both indoor and outdoor environments. The resulting workflow enhances the accurate detection of openings, facilitating the creation of BIM objects for integration into a 3D model. Evaluation results are promising, demonstrating the effectiveness of combining geometry and radiometry in point cloud analysis.