Skip to Main content Skip to Navigation
Theses

Détection de Changement en Imagerie Radar

Abstract : One of the problems encountered during the exploitation of large volumes of data whilst in temporal series relates to the selections of sets of pertinent images, enabling an estimation of good quality of the deformation signal. The quality of an interferometric dataset and the differential measures which are associated to them (coregistration, interferometric phase) can be determined a priori by a coherence model [Zebker et Villasenor, 1992]. In practice, the confrontation of this model to real observations is rendered difficult due to temporal variability in the studied sequences. At first, we focus on the qualification coherence model on 82 ERS-1 / 2 acquired between1992 and 2002. The confrontation of this model with experimental measurements on a temporally stable coherent area (dam of Serre-Ponçon, Hautes-Alpes) reveals loss of coherence due to inaccurate coregistration, thus requesting a method for a better estimation of the coherence. Secondly, a strategy for the selection of sets adapted to the global coregistration of this temporal series which lacked in coherence had been devised in order to better the estimation of the coherence. We express the distance between images using a triangular conorm, used to select the best set of image pairs using a minimum spanning tree, as it has been suggested by [Refice et al.,2006]. We characterize the limits of this approach and propose new solutions based on two novel methods for selecting image pairs including redundancy. The contribution of these methods is demonstrated by simulation and with ERS data, highlighting the trade-off between the number of image pairs treated and the quality of the overall result.
Complete list of metadatas

https://pastel.archives-ouvertes.fr/pastel-00597421
Contributor : Guillaume Hochard <>
Submitted on : Tuesday, May 31, 2011 - 6:24:30 PM
Last modification on : Friday, July 31, 2020 - 10:44:07 AM
Long-term archiving on: : Sunday, December 4, 2016 - 7:37:32 AM

Identifiers

  • HAL Id : pastel-00597421, version 1

Collections

Citation

Guillaume Hochard. Détection de Changement en Imagerie Radar. Traitement du signal et de l'image [eess.SP]. Télécom ParisTech, 2011. Français. ⟨pastel-00597421⟩

Share

Metrics

Record views

811

Files downloads

828