An Automated Approach to the Generation of Structured Building Information Models from Unstructured 3d Point Cloud Scans

In this paper we present and evaluate an approach for the automatic generation of building models in IFC BIM format from unstructured Point Cloud scans, as they result from 3dlaser scans of buildings.While the actual measurement process is relatively fast, 85% of the overall time are spend on the interpretation and transformation of the resulting Point Cloud data into information, which can be used in architectural and engineering design workflows. Our approach to tackle this problem, is in contrast to existing ones which work on the levels of points, based on the detection of building elements, such as walls, ceilings, doors, windows, and spaces and the relation between these. We present use cases with our software prototype, evaluate the results, and discuss future work, that will bring the research further towards the aim to create automatically semantic links between the conception of building design in BIM and simulations with the build environment.