From PC2BIM: Automatic Model generation from Indoor Point Cloud

In this paper, we present a system to automatically generate BIMs1 model from indoor point cloud. In contrary to previous works, our approach is able to take as input a point cloud with the minimum of information namely the points of coordinates (x, y, z) and produce excellent results. We first detect major flat surfaces such a walls, floor, and ceiling which are the bedrocks of our structure. Then, we present a novel 2D matrix template representation of walls which ease the operations like room layout and openings detection in polynomial time. Finally, we generate the BIM model rich with spatial and semantic information about the physical structures. A series of experiments performed show the efficiency and the precision of our approach.