Variational Image Motion Estimation by Accelerated Dual Optimization

Huaijiang Sun,Xue–Cheng Tai,Jing Yuan
DOI: https://doi.org/10.48550/arxiv.2011.10979
2020-01-01
Abstract:Estimating optical flows is one of the most interesting problems in computer vision, which estimates the essential information about pixel-wise displacements between two consecutive images. This work introduces an efficient dual optimization framework with accelerated preconditioners to the challenging nonsmooth optimization problem of total-variation regularized optical-flow estimation. In theory, the proposed dual optimization framework brings an elegant variational analysis on the given difficult optimization problem, while presenting an efficient algorithmic scheme without directly tackling the corresponding nonsmoothness in numeric. By introducing efficient preconditioners with a multi-scale implementation, the proposed accelerated dual optimization approaches achieve competitive estimation results of image motion, comparing to the state-of-the-art methods. Moreover, we show that the proposed preconditioners can guarantee convergence of the implemented numerical schemes with high efficiency.
What problem does this paper attempt to address?