Location Uncertainty Principle: Toward the Definition of Parameter-Free Motion Estimators

In this paper, we propose a novel optical flow approach for estimating two-dimensional velocity fields from an image sequence, which depicts the evolution of a passive scalar transported by a fluid flow. The Eulerian fluid flow velocity field is decomposed into two components: a large-scale motion field and a small-scale uncertainty component. We define the small-scale component as a random field. Then the data term of the optical flow formulation is based on a stochastic transport equation, derived from a location uncertainty principle [17]. In addition, a specific regularization term built from the assumption of constant kinetic energy involves the same diffusion tensor as the one appearing in the data transport term. This enables us to devise an optical flow method dedicated to fluid flows in which the regularization parameter has a clear physical interpretation and can be easily estimated. Experimental evaluations are presented on both synthetic and real images. Results indicate very good performance of the proposed parameter-free formulation for turbulent flow motion estimation.