Cahiers du CEREMADE

Unité Mixte de Recherche du C.N.R.S. N°7534
Abstract : This article proposes the use of manifolds to describe the non-local interactions of geometric structures in signals and images. We focus on the high dimensional set of patches extracted from signals and images. The manifold structure of such sets is detailed for various ensembles suitable for natural signals, images and textures modeling. These manifolds offer a low-dimensional parameterization of geometric patterns that can be found in such datasets. Diffusions over these manifolds are equivalent to recently proposed non-local processing methods. These algorithms are related to thresholding in an orthogonal basis adapted to the geometry of the underlying signal.
Manifold Models for Signals and Images
PEYRE Gabriel
Université de PARIS - DAUPHINE
Place du Maréchal de Lattre De Tassigny - 75775 PARIS CEDEX 16 - FRANCE
Téléphone : +33 (0)1 44-05-49-23 - fax : +33 (0)1 44-05-45-99