Skip to Main content Skip to Navigation

Outils géométriques pour la modélisation et la reconnaissance d'objets polyédriques

Abstract : This thesis is devoted to a procedure for constructing models of polyhedral objects with the goal of subsequently recognizing individual objects from single images. The problem consists of determining which information is necessary for recognition and how to extract it from a set of images representing the object to be modeled. Two kinds of information are studied. Firstly, as a part of projective geometry and invariant theory, we show how to extract a numerical 3D information (point coplanarity, cross ratios) characteristic of an objet from two matched images of it. Secondly, we propound a method to extract visual information from a set of images. These images are matched, their content is compared, and they are grouped according to the object aspect they represent. Each of the aspects gives rise to a model subpart which collate the available information. The different algorithms described in the thesis are tested with real and synthetic images.
Document type :
Complete list of metadatas
Contributor : Thèses Imag <>
Submitted on : Thursday, February 26, 2004 - 1:29:46 PM
Last modification on : Friday, November 6, 2020 - 3:54:02 AM
Long-term archiving on: : Wednesday, September 12, 2012 - 3:50:18 PM


  • HAL Id : tel-00005127, version 1




Patrick Gros. Outils géométriques pour la modélisation et la reconnaissance d'objets polyédriques. Interface homme-machine [cs.HC]. Institut National Polytechnique de Grenoble - INPG, 1993. Français. ⟨tel-00005127⟩



Record views


Files downloads