Variational Image Motion Estimation by Accelerated Dual Optimization

TL;DR

This paper presents an accelerated dual optimization framework with preconditioners for efficient and accurate optical flow estimation, improving convergence and performance over existing methods.

Contribution

It introduces a novel dual optimization approach with preconditioners for total-variation regularized optical flow, enabling efficient and convergent solutions without directly handling nonsmoothness.

Findings

Achieves competitive optical flow estimation results.

Guarantees convergence with high efficiency.

Outperforms some state-of-the-art methods.

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.