Automated Processes for Preparing a Point Cloud Within the Scan-To-BIM Methodology

The Architecture Engineering Construction (AEC) industry is increasingly considering and taking up Building Information Modelling (BIM), with different tricks and depending on development needs. Scan-to-BIM methodologies, which enable the creation of the as-built digital model starting from the point cloud collection and processing, which have been widely used recently, are the first reference step in the digital representation and reproduction of architectural and structural elements. In this context, a holistic method is proposed relying on the adoption of an efficient workflow for the arrangement of a point cloud, referring to a disused industrial building, for subsequent ensuing BIM authoring procedures. Throughout the model creation process, as well as segmentation techniques, were employed to ensure the preservation of significant geometric and semantic information. Based on reference height levels, it was decided to apply an initial splitting of the point cloud. After adopting different denoising techniques to remove unnecessary points, the items were semi-automatically grouped into separate point clouds using classification procedures, and model-fitting algorithms were utilised to find dominating planes connected to some analysed elements. The final stage, decimation, enabled the analysis of a point cloud that had been thoroughly cleaned up and streamlined in which the building elements deemed to be crucial to the modeling procedures had already been identified. This provides vetting in a multi-user and multi-disciplinary BIM authoring environment while maintaining a shared spatial reference.