Saisie photogrammétrique multi-représentation de bâtiments : une approche Semi-Automatisée Initialisée et Supportée par l'Intervention humainE

Authors: Frédéricque, Benoit
Advisor: Daniel, SylvieBédard, Yvan
Abstract: 3D MRDB (Multi Representation Data Base) population is more and more required to support advanced cartographical applications and advanced geospatial decisional analysis. This dissertation presents a new photogrammetric approach dedicated to multiple representation acquisition process to populate the buildings of a 3D MRDB. The proposed approach is named SAISIE (this French acronym matches with a semi-automatic acquisition process, initialized and supported by human intervention). The SAISIE approach tackles simultaneously the Detailed Geometries (DG) extraction and the Simplified Geometries (GS) extraction. This uses both the Multi-Representation Acquisition Pattern concept and the Instance Driven SASS concept (SASS : Selection of the Algorithms, Sources and Setting) to improve the process performance. These two new concepts have been introduced during this research. The MRAP concept stems from bridging together the geometric pattern concept (used to support generalisation process) and the parametric model (used to support the photogrammetric building extraction). Two new algorithms have also been introduced. The first one deals with the automatic implantation of 3D geometric pattern and the second one with the automatic extraction of building footprints. The SAISIE approach, the new concepts and the two new algorithms, have been implemented and tested with four test sites. These test sites cover more than three hundred buildings. Results analysis and several recommendations, based on our experimentation and experience, are proposed to conclude this dissertation.
Document Type: Thèse de doctorat
Issue Date: 2008
Open Access Date: 13 April 2018
Permalink: http://hdl.handle.net/20.500.11794/19747
Grantor: Université Laval
Collection:Thèses et mémoires

Files in this item:
SizeFormat 
25094.pdf10.18 MBAdobe PDFView/Open
All documents in CorpusUL are protected by Copyright Act of Canada.