Skip to Main content Skip to Navigation
New interface
Conference papers

Coupling reduced models for optimal motion estimation

Abstract : The paper discusses the issue of motion estimation by image assimilation in numerical models, based on Navier-Stokes equations. In such context, models' reduction is an attractive approach that is used to decrease cost in memory and computation time. A reduced model is obtained from a Galerkin projection on a subspace, defined by its orthogonal basis. Long temporal image sequences may then be processed by a sliding-window method. On the first sub-window, a fixed basis is considered to define the reduced model. On the next ones, a Principal Order Decomposition is applied, in order to define a basis that is simultaneously small-size and adapted to the studied image data. Results are given on synthetic data and quantified according to state-of-the-art methods. Application to satellite images demonstrates the potential of the approach.
Document type :
Conference papers
Complete list of metadata

Cited literature [7 references]  Display  Hide  Download
Contributor : Nathalie Gaudechoux Connect in order to contact the contributor
Submitted on : Friday, November 15, 2013 - 4:46:15 PM
Last modification on : Friday, January 21, 2022 - 3:19:50 AM
Long-term archiving on: : Sunday, February 16, 2014 - 4:28:25 AM


Files produced by the author(s)


  • HAL Id : hal-00803622, version 1



Karim Drifi, Isabelle Herlin. Coupling reduced models for optimal motion estimation. ICPR - 21st International Conference on Pattern Recognition, Nov 2012, Tsukuba, Japan. pp.2651-2654. ⟨hal-00803622⟩



Record views


Files downloads