Advanced Data Terms for Variational Optic Flow Estimation

In this paper, we present optic flow algorithms which are based on a variety of increasingly sophisticated data terms. Such data terms allow to better identify correspondences between points in either image than the traditional intensity difference since they characterize the local image structure more uniquely. We present an algorithmic framework which allows to directly incorporate arbitrary data terms: In contrast to traditional approaches the minimization scheme is not based on local linearization. In particular, we quantitatively compare the classical intensity similarity with nonconvex truncated data terms, with a patch-based intensity difference and with a patch-based normalized cross correlation. Experiments demonstrate that at the expense of additional runtime more advanced data terms may help to improve flow estimates.