Techniques de mise en correspondance et détection de changements

Abstract : We will present matching techniques for several kinds of objects. We fill first deal with geodesic matching of objects, which consists in defining a geodesic distance on the object set considered as a Riemannian manifold. This distance will have to take into account the action of a group of deformations on the object set and be invariant with respect to this group action. We will use this technique for the matching of images on which a group of diffeomorphisms acts. We will deduce an algorithm that will be improved by a multiresolution strategy. We will also use this geodesic distance to match features such as points or vectors thanks to a spline based modeling of the deformation. Finally, we will present some matching techniques for feature sets which allows a one to one pairing of the features.
Document type :
Theses
Liste complète des métadonnées

https://tel.archives-ouvertes.fr/tel-00133078
Contributor : Ens Cachan Bibliothèque <>
Submitted on : Friday, February 23, 2007 - 2:41:17 PM
Last modification on : Thursday, January 11, 2018 - 6:12:21 AM
Document(s) archivé(s) le : Wednesday, April 7, 2010 - 2:53:02 AM

Identifiers

  • HAL Id : tel-00133078, version 1

Collections

Citation

Laurent Garcin. Techniques de mise en correspondance et détection de changements. Mathématiques [math]. École normale supérieure de Cachan - ENS Cachan, 2005. Français. ⟨tel-00133078⟩

Share

Metrics

Record views

617

Files downloads

193