IMAGE RETRIEVAL FOR 3D MODELLING OF ARCHITECTURE USING AI AND PHOTOGRAMMETRY

Abstract. This research is intended to provide an initial solution to the problem of finding images for processing by photogrammetry in special cases where these do not exist. An overview of existing artificial intelligence-based algorithms that enable the extension of source image dataset is reported. In particular, this research focused on the use of prompt-to-image systems for obtaining images to be used in reconstruction and then in the next step of 3D modelling. Thus, the combined use of these three techniques, AI, photogrammetry, and modelling allowed the creation of a model of a building that never existed except in the collective imagination, which is the tower of Babel. In particular, the case study chosen is the illustration in Kircher book present in the library of the Brixen seminary that is closed to the public and for which it was necessary to create a tool to enhance the value and knowledge of this heritage for external users. Therefore, the creation of an augmented reality app enabled the visualization of the model created by offering possibilities for immersive experiences and dissemination of the research to a wide audience.