Occlusion Detection in Dense Stereo Estimation with Convex Optimization

Antonin Chambolle (1) , Pascal Monasse (2) , Pauline Tan (3)

1 CMAP - Centre de Mathématiques Appliquées
2 LIGM - Laboratoire d'Informatique Gaspard-Monge
3 ONERA - The French Aerospace Lab [Palaiseau]

In this paper, we propose a dense two-frame stereo algorithm which handles occlusion in a variational framework. Our method is based on a new regularization model which includes both a constraint on the occlusion width and a visibility constraint in nonoccluded areas. The minimization of the resulting energy functional is done by convex relaxation. A post-processing then detects and fills the occluded regions. We also propose a novel dissimilarity measure that combines color and gradient comparison with a variable respective weight, to benefit from the robustness of the comparison based on local variations while avoiding the fattening effect it may generate.

Keywords: Stereo-matching; occlusion detection; variational method; convex relaxation

Domains: Computer Vision and Pattern Recognition [cs.CV]

Antonin Chambolle, Pascal Monasse, Pauline Tan. Occlusion Detection in Dense Stereo Estimation with Convex Optimization. ICIP'17, IEEE International Conference on Image Processing, Sep 2017, Pekin, China. ⟨10.1109/ICIP.2017.8296741⟩. ⟨hal-01700678⟩