High-dimension multi-label problems: convex or non convex relaxation?
PAPADAKIS, Nicolas
Modelling, Observations, Identification for Environmental Sciences [MOISE]
Institut de Mathématiques de Bordeaux [IMB]
See more >
Modelling, Observations, Identification for Environmental Sciences [MOISE]
Institut de Mathématiques de Bordeaux [IMB]
PAPADAKIS, Nicolas
Modelling, Observations, Identification for Environmental Sciences [MOISE]
Institut de Mathématiques de Bordeaux [IMB]
< Reduce
Modelling, Observations, Identification for Environmental Sciences [MOISE]
Institut de Mathématiques de Bordeaux [IMB]
Language
en
Article de revue
This item was published in
SIAM Journal on Imaging Sciences. 2013-12-02, vol. 6, n° 4, p. 2603-2639
Society for Industrial and Applied Mathematics
English Abstract
This paper is concerned with the problem of relaxing non convex functionals, used in image processing, into convex problems. We review most of the recently introduced relaxation methods, and we propose a new convex one ...Read more >
This paper is concerned with the problem of relaxing non convex functionals, used in image processing, into convex problems. We review most of the recently introduced relaxation methods, and we propose a new convex one based on a probabilistic approach, which has the advantages of being intuitive, flexible and involving an algorithm without inner loops. We investigate in detail the connections between the solutions of the relaxed functionals with a minimizer of the original one. Such connection is demonstrated only for non convex relaxation which turns out to be quite robust to initialization. As a case of study, we illustrate our theoretical analysis with numerical experiments, namely for the optical flow problem.Read less <
English Keywords
multilabel problems
convex relaxation
segmentation
disparity and optical flow
ANR Project
Adaptivité pour la représentation des images naturelles et des textures - ANR-08-EMER-0009
Origin
Hal imported