Determining occlusions from space and time image reconstructions

Abstract : The problem of localizing occlusions between consecutive frames of a video is important but rarely tackled on its own. In most works, it is tightly interleaved with the computation of accurate optical flows, which leads to a delicate chicken-and-egg problem. With this in mind, we propose a novel approach to occlusion detection where visibility or not of a point in next frame is formulated in terms of visual reconstruction. The key issue is now to determine how well a pixel in the first image can be " reconstructed " from co-located colors in the next image. We first exploit this reasoning at the pixel level with a new detection criterion. Contrary to the ubiquitous displaced-frame-difference and forward-backward flow vector matching, the proposed alternative does not critically depend on a pre-computed, dense displacement field, while being shown to be more effective. We then leverage this local modeling within an energy-minimization framework that delivers oc-clusion maps. An easy-to-obtain collection of parametric motion models is exploited within the energy to provide the required level of motion information. Our approach outper-forms state-of-the-art detection methods on the challenging MPI Sintel dataset.
Document type :
Conference papers
Complete list of metadatas

Cited literature [42 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01307703
Contributor : Juan-Manuel Perez-Rua <>
Submitted on : Tuesday, April 26, 2016 - 4:25:02 PM
Last modification on : Wednesday, April 11, 2018 - 1:55:08 AM
Long-term archiving on : Tuesday, November 15, 2016 - 1:51:53 PM

File

occlusions_cvpr2016.pdf
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution - NonCommercial - ShareAlike 4.0 International License

Identifiers

  • HAL Id : hal-01307703, version 1

Collections

Citation

Juan-Manuel Pérez-Rúa, Tomas Crivelli, Patrick Bouthemy, Patrick Pérez. Determining occlusions from space and time image reconstructions. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , Jun 2016, Las Vegas, United States. ⟨hal-01307703⟩

Share

Metrics

Record views

1980

Files downloads

842