Skip to Main content Skip to Navigation

Estimation du mouvement par assimilation de données dans des modèles dynamiques d'ordre réduit

Abstract : Motion estimation is a major challenge in the field of image sequence analysis. This thesis is a study of the dynamics of geophysical flows visualized by satellite imagery. Satellite image sequences are currently underused for the task of motion estimation. A good understanding of geophysical flows allows a better analysis and forecast of phenomena in domains such as oceanography and meteorology. Data assimilation provides an excellent framework for achieving a compromise between heterogeneous data, especially numerical models and observations. Hence, in this thesis we set out to apply variational data assimilation methods to estimate motion on image sequences. As one of the major drawbacks of applying these assimilation techniques is the considerable computation time and memory required, we therefore define and use a model reduction method in order to significantly decrease the necessary computation time and the memory. We then explore the possibilities that reduced models provide for motion estimation, particularly the possibility of strictly imposing some known constraints on the computed solutions. In particular, we show how to estimate a divergence free motion with boundary conditions on a complex spatial domain.
Document type :
Complete list of metadata

Cited literature [84 references]  Display  Hide  Download
Contributor : Nathalie Gaudechoux Connect in order to contact the contributor
Submitted on : Friday, July 19, 2013 - 4:21:28 PM
Last modification on : Friday, January 21, 2022 - 3:21:35 AM
Long-term archiving on: : Monday, October 21, 2013 - 11:05:09 AM


  • HAL Id : tel-00846688, version 1


Karim Drifi. Estimation du mouvement par assimilation de données dans des modèles dynamiques d'ordre réduit. Modélisation et simulation. Université Pierre et Marie Curie - Paris VI, 2013. Français. ⟨tel-00846688⟩



Record views


Files downloads