Methods to process low-level CAD plans and creative Building Information Models (BIM)

Abstract

This dissertation goes in depth to the detection of semantic elements from CAD building floor plans in order to create Building Information Models (BIM). The dissertation can be divided into two main parts. In the first part we describe a number of algorithms which allow obtaining semiautomatically semantic elements from CAD architectural floor plans: walls, doors and windows, staircases, closed rooms. In the second part, we study the management of the topological information of the semantic elements which were detected by using the algorithms described in the first part. We propose a topologically correct model which includes 2D information (topology graph) and 3D information (obtained using an algorithm called triple extrusion). The information of this model combines geometric, topological and semantic elements, and allow obtaining CityGML models.