资源论文Dense, Accurate Optical Flow Estimation with Piecewise Parametric Model

Dense, Accurate Optical Flow Estimation with Piecewise Parametric Model

2019-12-25 | |  62 |   37 |   0

Abstract

This paper proposes a simple method for estimatingdense and accurate optical flow field. It revitalizes an earlyidea of piecewise parametric flow model. A key innovation is that, we fit a flow field piecewise to a variety of para-metric models, where the domain of each piece (i.e., eachpiece’s shape, position and size) is determined adaptively,while at the same time maintaining a global inter-piece flowcontinuity constraint. We achieve this by a multi-model fitting scheme via energy minimization. Our energy takesinto account both the piecewise constant model assumptionand the flow field continuity constraint, enabling the pro-posed method to effectively handle both homogeneous mo-tions and complex motions. The experiments on three public optical flow benchmarks (KITTI, MPI Sintel, and Middlebury) show the superiority of our method compared with the state of the art: it achieves top-tier performances on all the three benchmarks.

上一篇:A Graphical Model Approach for Matching Partial Signatures

下一篇:Combination Features and Models for Human Detection

用户评价
全部评价

热门资源

  • Learning to Predi...

    Much of model-based reinforcement learning invo...

  • Stratified Strate...

    In this paper we introduce Stratified Strategy ...

  • The Variational S...

    Unlike traditional images which do not offer in...

  • A Mathematical Mo...

    Direct democracy, where each voter casts one vo...

  • Rating-Boosted La...

    The performance of a recommendation system reli...