Fully Trainable and Interpretable Non-Local Sparse Models for Image Restoration - Inria - Institut national de recherche en sciences et technologies du numérique
Conference Papers Year : 2020

Fully Trainable and Interpretable Non-Local Sparse Models for Image Restoration

Abstract

Non-local self-similarity and sparsity principles have proven to be powerful priors for natural image modeling. We propose a novel differentiable relaxation of joint sparsity that exploits both principles and leads to a general framework for image restoration which is (1) trainable end to end, (2) fully interpretable, and (3) much more compact than competing deep learning architectures. We apply this approach to denoising, jpeg deblocking, and demosaicking, and show that, with as few as 100K parameters, its performance on several standard benchmarks is on par or better than state-of-the-art methods that may have an order of magnitude or more parameters.
Fichier principal
Vignette du fichier
paper.pdf (6.86 Mo) Télécharger le fichier
Origin Files produced by the author(s)
Loading...

Dates and versions

hal-02414291 , version 1 (16-12-2019)
hal-02414291 , version 2 (24-03-2020)
hal-02414291 , version 3 (24-07-2020)

Identifiers

Cite

Bruno Lecouat, Jean Ponce, Julien Mairal. Fully Trainable and Interpretable Non-Local Sparse Models for Image Restoration. ECCV 2020 - European Conference on Computer Vision, Aug 2020, Glasgow / Virtual, United Kingdom. pp.238-254, ⟨10.1007/978-3-030-58542-6_15⟩. ⟨hal-02414291v3⟩
540 View
696 Download

Altmetric

Share

More