3D MODELING OF INDUSTRIAL HERITAGE BUILDING USING COTSs SYSTEM: TEST, LIMITS AND PERFORMANCES
- 1Department of Environmental, Land and Infrastructure Engineering (DIATI), Polytechnic of Torino corso Duca degli Abruzzi 24, 10129 Torino, Italy
- 2Polytechnic Department of Engineering and Architecture (DPIA), University of Udine via delle Scienze 206, 33100 Udine, Italy
Keywords: Benchmark session, UAV photogrammetry, camera “on-the-job” calibration, oblique images, COTS
Abstract. The role of UAV systems in applied geomatics is continuously increasing in several applications as inspection, surveying and geospatial data. This evolution is mainly due to two factors: new technologies and new algorithms for data processing. About technologies, from some years ago there is a very wide use of commercial UAV even COTSs (Commercial On-The-Shelf) systems. Moreover, these UAVs allow to easily acquire oblique images, giving the possibility to overcome the limitations of the nadir approach related to the field of view and occlusions. In order to test potential and issue of COTSs systems, the Italian Society of Photogrammetry and Topography (SIFET) has organised the SBM2017, which is a benchmark where all people can participate in a shared experience. This benchmark, called “Photogrammetry with oblique images from UAV: potentialities and challenges”, permits to collect considerations from the users, highlight the potential of these systems, define the critical aspects and the technological challenges and compare distinct approaches and software. The case study is the “Fornace Penna” in Scicli (Ragusa, Italy), an inaccessible monument of industrial architecture from the early 1900s. The datasets (images and video) have been acquired from three different UAVs system: Parrot Bebop 2, DJI Phantom 4 and Flytop Flynovex. The aim of this benchmark is to generate the 3D model of the “Fornace Penna”, making an analysis considering different software, imaging geometry and processing strategies. This paper describes the surveying strategies, the methodologies and five different photogrammetric obtained results (sensor calibration, external orientation, dense point cloud and two orthophotos), using separately – the single images and the frames extracted from the video – acquired with the DJI system.