Désoccultation de cartes de profondeurs pour la synthèse de vues virtuelles

Pierre Buyssens 1 Olivier Le Meur 1 Maxime Daisy 2 David Tschumperlé 2 Olivier Lézoray 2
1 Sirocco - Analysis representation, compression and communication of visual data
Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
2 Equipe Image - Laboratoire GREYC - UMR6072
GREYC - Groupe de Recherche en Informatique, Image, Automatique et Instrumentation de Caen
Abstract : In this paper, we propose to tackle the specific problem of the disocclusion of depth-maps when synthesizing virtual views from RGB-D images. The warping process used to synthesize such a new view creates indeed holes in the depth-map that we propose to inpaint. A detailed analysis of the depth-map disocclusion problem leads us to the proposal of an intuitive algorithm. In particular, our approach recovers well lost structures of the objects, which is essential for the visual consistency of the final virtual view. Comparisons of our approach to state-of-the-art methods confirm the relevance of our proposal.
Document type :
Conference papers
Complete list of metadatas

Cited literature [13 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01320967
Contributor : Pierre Buyssens <>
Submitted on : Tuesday, May 24, 2016 - 3:54:44 PM
Last modification on : Thursday, November 14, 2019 - 11:53:52 AM

File

rfia_depth_buyssens.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01320967, version 1

Citation

Pierre Buyssens, Olivier Le Meur, Maxime Daisy, David Tschumperlé, Olivier Lézoray. Désoccultation de cartes de profondeurs pour la synthèse de vues virtuelles. RFIA 2016, Jun 2016, Clermont-Ferrand, France. ⟨hal-01320967⟩

Share

Metrics

Record views

405

Files downloads

136