Skip to Main content Skip to Navigation

Détection et suivi de structures déformables en mouvement, application à la météorologie

Abstract : The purpose of this thesis is the detection of deformable structures and the estimation of their motion. This work is applied to satellite imagery, and specifically focused on meteorological infrared images. On these images, the structures of interest are clouds. A markovian model is proposed to locate these structures. This model captures cloud structures with homogeneity and regularity properties. The second part of this work introduces a novel method for motion estimation. The classical hypothesis underlying motion estimation is the invariance of brightness of each point during its displacement. This hypothesis is not valid in the context of our work. We therefore propose a method based on the invariance of moving objects' total brightness. Applied to clouds on infrared data, this hypothesis is proved to be equivalent to the invariance of volume. We demonstrate that this leads to a constraint strictly equivalent to the equation of mass conservation in fluid mechanisms. This partial differential equation is then solved within a variational framework. Two alternatives are proposed, the first one using a prior affine motion model, the second unifying the two invariance principles mentioned above. These methods are applied to the reconstruction of clouds' trajectories. A framework is proposed to analyse these trajectories: it is based on a decomposition of trajectories on a spline functions basis, this decomposition being plugged into a data analysis system
Document type :
Complete list of metadata
Contributor : Nathalie Gaudechoux Connect in order to contact the contributor
Submitted on : Thursday, June 23, 2016 - 10:30:54 AM
Last modification on : Friday, February 4, 2022 - 3:10:10 AM
Long-term archiving on: : Saturday, September 24, 2016 - 11:23:58 AM


  • HAL Id : tel-01336443, version 1



Dominique Béréziat. Détection et suivi de structures déformables en mouvement, application à la météorologie. Modélisation et simulation. Université de Paris-Sud, 1999. Français. ⟨tel-01336443⟩



Record views


Files downloads