This brief focuses on two main problems in the domain of optical flow
and trajectory estimation: (i) The problem of finding convex
optimization methods to apply sparsity to optical flow; and (ii) The
problem of how to extend sparsity to improve trajectories in a
computationally tractable way.
Beginning with a review of optical flow fundamentals, it discusses the
commonly used flow estimation strategies and the advantages or
shortcomings of each. The brief also introduces the concepts associated
with sparsity including dictionaries and low rank matrices. Next, it
provides context for optical flow and trajectory methods including
algorithms, data sets, and performance measurement. The second half of
the brief covers sparse regularization of total variation optical flow
and robust low rank trajectories. The authors 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.
The brief is targeted at researchers and practitioners in the fields of
engineering and computer science. It caters particularly to new
researchers looking for cutting edge topics in optical flow as well as
veterans of optical flow wishing to learn of the latest advances in
multi-frame methods.