Computing optical flow via variational techniques

Defined as the apparent motion in a sequence of images, the optical flow is very important in the computer vision community where its accurate estimation is necessary for many applications. It is one of the most studied problems in computer vision. In spite of this, not much theoretical analysis has been done. In this article, we first present a review of existing variational methods. Then, we will propose an extended model that will be rigorously justified on the space of functions of bounded variations. Finally, we present an algorithm whose convergence will be carefully demonstrated. Some results showing the capabilities of this method conclude the paper.