Recognition and Positioning of SBCs in BIM Models Using a Geometric vs Colour Consensus Approach

摘要

This paper proposes a new approach that recognizes different kinds of small objects that are commonly seen on walls and ceilings of buildings (Figure 1). The general idea supported in this paper is that such objects, called service building components (SBCs), should be inserted into the structure of the standard existing as-is BIM models, as they are automatically generated. But identifying these small building components within the vast dataset provided by the scanner is a difficult issue not solved yet. Our approach first processes a dense coloured point cloud and extracts global geometry and colour features to identify potential SBCs. A novel consensus algorithm between geometry and colour features is the basis to efficiently recognize and calculate the position of the existing small objects in the as-is BIM model of the building. The method has been tested in real and simulated environments providing encouraging results.