Processing existing building geometry for reuse as Linked Data

The Web currently hosts a vast amount of 2D images and 3D building models. Each repository has its own data structure and a limited set of semantics according to their own needs. With the advent of Semantic Web Technologies, the opportunity arises to combine these heterogeneous data sets and publish them as Linked Data. It is within the scope of this research to investigate whether online 2D and 3D content can be enriched, published and reused as RDF. The emphasis of this work is on extracting building component information from online building geometry and publishing it as Linked Data. An interpretation framework is presented that takes as input any building mesh and computes its building components through machine learning techniques. Additionally, a Structure-from-Motion pipeline is proposed that provides similar outputs and links the 2D imagery to the reconstructed 3D building geometry. The experiments show that, even though the building content originates from different sources and was not modeled according to any standards, building geometry in online repositories and photogrammetric reconstructions can be semantically enriched with component information using terminology from Linked Building Data ontologies such as BOT, PRODUCT and OMG/FOG/GOM. This is an important step towards making structureless geometric information retrievable, linkable and thus reusable over the Web.

Publication year:2020

BOF-keylabel:yes

IOF-keylabel:yes

BOF-publication weight:6

CSS-citation score:1

Authors from:Higher Education

Accessibility:Open