Scan-to-BIM works have so far mainly focused on 'structural' components such as floors, ceiling, walls (with doors and windows). But, the control of new facilities and the production of their corresponding as-is BIM models requires the identification and inspection of numerous other building components and objects, e.g. MEP components such as plugs, switches, ducts, and signs. In this paper, we present a novel 6D-based (XYZ + RGB) approach that processes dense coloured 3D points provided by terrestrial laser scanners to recognize such smaller objects that are commonly located on walls. This paper focuses on the recognition of objects such as sockets, switches, signs, and extinguishers. After segmenting the point clouds corresponding to the walls of a building, a set of candidate objects are detected independently in the colour and geometric spaces, and a consensus procedure integrates both results to infer recognition. The method has been tested on real indoors yielding promising results.
|Title of host publication||LC3 2017|
|Subtitle of host publication||Volume 1- Proceedings of the Joint Conference on Computing in Construction|
|Editors||Frédéric Bosché, Ioannis Brilakis, Rafael Sacks|
|Number of pages||8|
|Publication status||Published - Jul 2017|