3D Generalisation of Building Components – An Initial Proof of Concept

Abstract. A varied range of applications make use of 3D models nowadays, for instance in urban planning, energy demand studies, solar irradiation, or noise estimation. Acquisition, maintenance, and production of 3D spatial data is costly and laborious, especially at a national level, a great challenge for National Mapping and Cadastral Agencies (NMCAs) – such as Ordnance Survey (OS) for Great Britain. Generalisation is designed to address this challenge, where new datasets are created from a single source by the selection of the desired information and reduction of the amount of detail and data volume. Extensive literature exists in the context of 2D generalisation and automated algorithms exist to remove unwanted detail, however, adding a third dimension complicates the process significantly. Here, a methodology to address this issue is proposed, where the façades of a 3D building are decomposed, rotated, and translated from 3D environment to 2D. Existing automated 2D generalisation operators are applied to building elements and once generalised, they are rotated back to 3D. The outer shell of the resulting generalised 3D building is reconstructed with the independently generalised façade. The results demonstrate a potential flexible, component-based method for 3D generalisation, that could benefit NMCAs.