Robust optical flow estimation for illumination changes

How to tackle the difficulty involved in optical flow calculation caused by illumination inconsistency in adjacent video frames remains an open challenge. This motivates us to investigate how to obtain a valid and accurate optical flow field which is robust to illumination change based on structure-texture decomposition. In this paper, we attempt to reveal the involved technique details in this ROF model based image decomposition and carefully examine the improvement extent quantitatively by experiment. The advent of structure-texture decomposition for optical flow calculation is verified by experiment results. It is demonstrated from the comparison in the experiment that the accuracy of optical flow field estimated by TV-L¹ model and structure-texture decomposition based method outperforms that obtained by pure TV-L¹ model and TV-L¹ and filter constancy based method.