Semantic interpretation of architectural and archaeological geometries: Point cloud segmentation for HBIM parameterisation

摘要

The creation of Cultural Heritage (CH) Digital Twins is based on i) the capture of geometric data using digital technologies (laser scanning and photogrammetry); ii) the processing of the raw data to identify, segment and label the objects; and iii) their conversion into BIM objects. Hitherto, the most extensive method for BIM seg-mentation and modelling is manual, which led to research into the automation of this process, also in the field of CH. Manual operations are still labour-intensive and mathematical approaches are not inclusive for all CH specialists. In this context, this research studies the application of Brodu and Lague's morphological segmen-tation algorithm called CANUPO to classify the architectural components of the facade of the 16th-century Casa de Pilatos Palace in Seville, Spain, from a Terrestrial Laser Scanning (TLS) point cloud dataset. In this paper, the experimentation on semantic segmentation was carried out using open-source software, specifically the CANUPO algorithm integrated into CloudCompare software.