FLUID

Workpackage-2 : Physical models for 2D and 3D fluid motion computation

Description

Physical models for 2D and 3D fluid motion computation. This workpackage adresses issues of 2D and 3D motion estimation from fluid-flow image sequences. This comprises tasks of local motion estimation through efficient filter banks and of global motion estimation. The last issue involves design of appropriate priors or studies of efficient vector field decompositions. All these tasks will be followed by taking into account tractable physical models and constraints. Furthermore, this workpackage will address the study of efficient numerical multiscale schemes.

Some demonstrations of realized tasks

1. MOTION ANALYSIS / FLUID FLOWS, A. Cuzol, E.Memin
   Proposition of a low dimensional motion estimator for fluid flows, relying on a smooth discretization of the vorticity and divergence maps. (DEMO HERE)
   
   
2. CONSISTENT SPATIO-TEMPORAL ESTIMATION OF STRATIFIED ATMOSPHERIC MOTION, P. Heas, E.Memin, N. Papadakis
   Proposition of a sound energy-based estimator producing time consistent dense motion fields related to the different layers in a stratified atmosphere. (DEMO HERE)
   
   
3. A COLLABORATIVE METHOD FOR DENSE MOTION ESTIMATION OF FLUID FLOWS, P. Heas, E.Memin, A. Szantai
   Proposition of a initialization scheme based on correlation-based velocity vectors for the dense estimation of fluid flows. The multiresolution scheme is avoided. (DEMO HERE)
   
   
4. ORTHOGONAL DECOMPOSITION AND ESTIMATION OF MOTION VECTOR FIELDS, J. Yuan, P. Ruhnau, E. Memin, C. Schnoerr
   (DEMO HERE)
   
   
5. TRACKING OF ATMOSPHERIC LAYER MOTION AND PRESSURE N. Papadakis, P. Heas, E.Memin
   Variational assimilation techniques applied to estimation and tracking of motion and pressure related to atmospheric layers. (DEMO HERE)
   
   
   

Involved partners

• Mannheim Univ.
• INRIA
• ULPGC
• CEMAGREF
• La Vision
• LMD-CNRS