Sparse regularization of TV-L1 optical flow

Abstract

Optical flow is an ill-posed underconstrained inverse problem. Many recent approaches use total variation (TV) to constrain the flow solution to satisfy color constancy. In this chapter we show that learning a 2-D overcomplete dictionary from the total variation result and then enforcing a sparse constraint on the flow improves the result. We describe a new approach that uses partially-overlapping patches to accelerate the calculation and is implemented in a coarse-to-fine strategy. Experimental results show that combining total variation and a sparse constraint from a learned dictionary is more effective than employing total variation alone.