Dense Fluid Flow Estimation
Résumé
In this paper we address the problem of estimating and analyzing the motion in image sequences showing fluid phenomenon. Due to the great deal of spatial and temporal distortions that luminance patterns exhibit in images of fluid, standard techniques from Computer Vision, originally designed for quasi-rigid motions with stable salient features, are not well adapted in this context. In that prospect, we investigate a dedicated energy-based motion estimator. The considered functional includes an original data model relying on the continuity equation of fluid mechanics. This new data model, which is specifically designed to be embedded in a multiresolution framework, is associated to an original div-curl type regularization. The optimization of the global energy function is solved within an efficient multigrid scheme. The performances of the resulting fluid flow estimator are demonstrated both on synthetic and real (meteorological) image sequences.