Recursive Optical Flow Estimation - Adaptive Filtering Approach

This paper presents a new approach towards the problem of recursive optical flow estimation from image sequences based on the differential framework proposed by Horn & Schunck. We show that gain is achieved both from computational and accuracy points of view when treating the estimation task recursively. Incorporation of the temporal axis into the estimation process is done by combining a temporal smoothness assumption on the optical flow. RLS and LMS recursive optical flow estimation algorithms are derived and tested.