HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation

Apport de la photogrammétrie et de l’intelligence artificielle à la détection des zones amiantées sur les fronts rocheux

Abstract : Recent regulation imposes to identify asbestos-bearing material before undertaking civil engineering works, both in existing buildings and in the natural environment. The purpose of the thesis project is to focus on the local scale of a rock outcrop in order to provide a 3D map of asbestos zones from photographs of the sites. In its natural context, asbestos occurrence is present on the surface of fractures with specific orientations. Three lines of research were followed. They are based on the processing of dense 3D point clouds obtained by photogrammetry.The first line of research focused on the spatial location and the orientation and frequency characterization of areas with a high density of rock outcrop fractures. The second concentrated on optimizing the shots to restore a rock fractured outcrop using the photogrammetry technique. The delineation of the asbestos zones in the (2D) photos was the starting point of a third line of research. This delineation was done manually in a first phase; the link between the points of a 3D model restored by photogrammetry and the pixels of the photos used for its 3D restitution allowed a 3D restitution of known asbestos zones, because they were identified in situ. The mapping was then extended to asbestos zones (that had not been identified in situ) by deep learning. A methodology integrating an auto-encoder (e.g. U-Net) has been developed to detect asbestos zones in 2D photos. The 2D-3D link made possible by the 3D photogrammetric restitution, rendered a 3D mapping of the asbestos zones.
Complete list of metadata

Contributor : Abes Star :  Contact
Submitted on : Thursday, March 10, 2022 - 12:01:10 PM
Last modification on : Saturday, March 12, 2022 - 3:14:09 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03512327, version 2



Philippe Caudal. Apport de la photogrammétrie et de l’intelligence artificielle à la détection des zones amiantées sur les fronts rocheux. Géologie appliquée. Le Mans Université, 2021. Français. ⟨NNT : 2021LEMA1030⟩. ⟨tel-03512327v2⟩



Record views


Files downloads