3D Data Processing: Towards the Automated BIM in Inhabited Indoors

In this paper we present a method for the reconstruction of interiors using a set of panoramic range data in scenes with clutter and occlusion. Although the ultimate goal of our project is to obtain automated Building Information Models of facilities, here we specifically deal with the reconstruction of simply-shaped wide areas (such as walls, ceilings and floors) behind furniture and facility pieces in interiors. Our approach is based on a sequential updating labeling strategy in different data representation spaces. A volumetric representation is used to permit the labeling of the 3D space for different range data and the fusion of all the scene's labels to obtain one single 2D labeling image for each of the simply-shaped wide areas of the room. Based on this labeling process, our method is able both to identify the occluded regions in the wall and, through a learning technique, to infer essential parts, such as doors, windows and closets. This method was tested under difficult clutter and occlusion conditions, yielding promising results. Scans were obtained using a state of the art laser scanner operated by a professional 3D scanning service provider.