Sparse Aggregation Framework for Optical Flow Estimation

Fortun, Denis; Bouthémy, Patrick; Kervrann, Charles

doi:10.1007/978-3-319-18461-6_26

Cited by 2 publications

(2 citation statements)

References 27 publications

(59 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We show that the method presented in this paper is faster and more robust to suboptimal candidate sets, while being competitive in terms of quantitative error. A first shorter version of this work was described in [37]. Compared to [37], we have integrated an occlusion handling module in the candidates estimation stage, we have modified the aggregation model to enforce the selection of a single candidate; we have improved the optimization step of our method, and we have extended the experimental validation of the method.…”

Section: Our Contributionsmentioning

confidence: 99%

See 1 more Smart Citation

A Variational Aggregation Framework for Patch-Based Optical Flow Estimation

2016

Self Cite

View full text Add to dashboard Cite

We propose a variational aggregation method for optical flow estimation. It consists of a two-step framework, first estimating a collection of parametric motion models to generate motion candidates, and then reconstructing a global dense motion field. The aggregation step is designed as a motion reconstruction problem from spatially varying sets of motion candidates given by parametric motion models. Our method is designed to capture large displacements in a variational framework without requiring any coarse-tofine strategy. We handle occlusion with a motion inpainting approach in the candidates computation step. By performing parametric motion estimation, we combine the robustness to noise of local parametric methods with the accuracy yielded by global regularization. We demonstrate the performance of our aggregation approach by comparing it to standard variational methods and a discrete aggregation approach on the Middlebury and MPI Sintel datasets.

show abstract

Section: Our Contributionsmentioning

confidence: 99%

“…In [37], we proposed a related model in a sparse representation framework, where the number of selected candidates was controlled by a sparsity constraint on α. The confidence measures were associated to the sparsity constraint with a weighted ℓ 1 penalization function.…”

Section: M(x)mentioning

confidence: 99%