Automatic facade modelling using point cloud data for energy-efficient retrofitting

Energy-efficient retrofitting of existing buildings is a key aspect for reaching the proposed energy consumption reduction targets fixed by public authorities in different countries. For this task, the availability of as-built building models is of primary importance for both diagnosis of thermal dispersion and designing of retrofitting. In this paper, we present an automated methodology to derive highly detailed 3D vector models of existing building facades starting from terrestrial laser scanning data. The presented methodology first accomplishes the segmentation of the point cloud of a building facade into its planar elements. Then, starting from the identified planar clusters, facade breaklines are automatically extracted to be used later to generate a 3D vector model. During this final step, some priors on urban scenes like the prevalence of straight lines and orthogonal intersections are exploited to set additional constraints. The final product is a semantically enriched 3D model of the building facade that can be integrated in Building Information Model (BIM) for planned maintenance. Eventually, the integration between derived facade models and infrared thermography (IRT) is presented for energy efficiency evaluation of buildings and detection of thermal anomalies.