Automatic segmentation and classification of BIM elements from point clouds

Abstract

Laser techniques are widely used to perform topographic building surveys by providing massive information and point clouds comprised of millions of points in seconds. Point clouds allow the creation of 3D models that represent information of significant importance to the AEC/FM (Architecture, Engineering, Construction, and Facilities Management) domain. However, few tools exist related to the automatic modelling of point clouds. We present a method to automatically segment, classify, and model point clouds that were tested with two point clouds acquired via static and dynamic laser techniques. This approach generated accurate 3D surfaces of building elements, including floors, ceilings, walls columns, and content. A future study will involve transferring the 3D surfaces into Building Information Model elements.