Skip to Main content Skip to Navigation

Développement de méthodes pour la détection de changement

Abstract : The detection of the changes in a landscape is the implementation of techniques that are meant to locate, highlight and finally, understand its temporal evolution. Such techniques are frequently applied to satellite observations in numerous and various environmental applications. However, most of these applications are not based upon a general methodology that justifies or generalises the different technical solutions. The first objective of the thesis is to propose a framework and a general methodology for detection of change in remote sensing. Some definitions are established, which express landscape properties in terms of characteristic scales and observability. After having performed a bibliographical analysis, we propose a methodology that breaks down change detection applications into five stages: identification and characterisation of the different sources of effective and exogenous changes; geometric and radiometric data alignment; extraction of relevant information according to the changes under interest; creation of residual with respect to a temporal evolution model; decision-making and synthesis of the results. The analysis of the relationship between the effective changes and those actually observed reveals fundamental links with the properties of the remote sensing systems. The study of these links is the second objective of the thesis. The crucial role of data alignment and particularly of geometric alignment is highlighted. An innovative iterative geometric co-registration method was developed with enhanced performances. It starts from an existing algorithm and is based on a multi-resolution analysis of the images. It makes use of techniques of sup-pixel matching and local geometric deformation models. An innovative validation protocol based on simulated images is also proposed in order to assess both the quality of this method and its benefits compared to standard co-registration methods.
Document type :
Complete list of metadata

Cited literature [92 references]  Display  Hide  Download
Contributor : Magalie Prudon Connect in order to contact the contributor
Submitted on : Wednesday, April 28, 2010 - 10:19:46 AM
Last modification on : Wednesday, November 17, 2021 - 12:30:15 PM
Long-term archiving on: : Tuesday, September 28, 2010 - 1:20:07 PM


  • HAL Id : tel-00477115, version 1


Philippe Blanc. Développement de méthodes pour la détection de changement. Sciences de la Terre. École Nationale Supérieure des Mines de Paris, 1999. Français. ⟨NNT : 1999ENMP0910⟩. ⟨tel-00477115⟩



Record views


Files downloads