Non-local recurrent network for image restoration

Ding Liu, Bihan Wen, Yuchen Fan, Chen Change Loy, Thomas S Huang

Research output: Contribution to journalConference article

Abstract

Many classic methods have shown non-local self-similarity in natural images to be an effective prior for image restoration. However, it remains unclear and challenging to make use of this intrinsic property via deep networks. In this paper, we propose a non-local recurrent network (NLRN) as the first attempt to incorporate non-local operations into a recurrent neural network (RNN) for image restoration. The main contributions of this work are: (1) Unlike existing methods that measure self-similarity in an isolated manner, the proposed non-local module can be flexibly integrated into existing deep networks for end-to-end training to capture deep feature correlation between each location and its neighborhood. (2) We fully employ the RNN structure for its parameter efficiency and allow deep feature correlation to be propagated along adjacent recurrent states. This new design boosts robustness against inaccurate correlation estimation due to severely degraded images. (3) We show that it is essential to maintain a confined neighborhood for computing deep feature correlation given degraded images. This is in contrast to existing practice [41] that deploys the whole image. Extensive experiments on both image denoising and super-resolution tasks are conducted. Thanks to the recurrent non-local operations and correlation propagation, the proposed NLRN achieves superior results to state-of-the-art methods with many fewer parameters. The code is available at https://github.com/Ding-Liu/NLRN.

Original languageEnglish (US)
Pages (from-to)1673-1682
Number of pages10
JournalAdvances in Neural Information Processing Systems
Volume2018-December
StatePublished - Jan 1 2018
Event32nd Conference on Neural Information Processing Systems, NeurIPS 2018 - Montreal, Canada
Duration: Dec 2 2018Dec 8 2018

Fingerprint

Recurrent neural networks
Image reconstruction
Image denoising
Experiments

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Information Systems
  • Signal Processing

Cite this

Liu, D., Wen, B., Fan, Y., Loy, C. C., & Huang, T. S. (2018). Non-local recurrent network for image restoration. Advances in Neural Information Processing Systems, 2018-December, 1673-1682.

Non-local recurrent network for image restoration. / Liu, Ding; Wen, Bihan; Fan, Yuchen; Loy, Chen Change; Huang, Thomas S.

In: Advances in Neural Information Processing Systems, Vol. 2018-December, 01.01.2018, p. 1673-1682.

Research output: Contribution to journalConference article

Liu, D, Wen, B, Fan, Y, Loy, CC & Huang, TS 2018, 'Non-local recurrent network for image restoration', Advances in Neural Information Processing Systems, vol. 2018-December, pp. 1673-1682.
Liu, Ding ; Wen, Bihan ; Fan, Yuchen ; Loy, Chen Change ; Huang, Thomas S. / Non-local recurrent network for image restoration. In: Advances in Neural Information Processing Systems. 2018 ; Vol. 2018-December. pp. 1673-1682.
@article{895bb76f6e474267b79d6a5ffb7da47c,
title = "Non-local recurrent network for image restoration",
abstract = "Many classic methods have shown non-local self-similarity in natural images to be an effective prior for image restoration. However, it remains unclear and challenging to make use of this intrinsic property via deep networks. In this paper, we propose a non-local recurrent network (NLRN) as the first attempt to incorporate non-local operations into a recurrent neural network (RNN) for image restoration. The main contributions of this work are: (1) Unlike existing methods that measure self-similarity in an isolated manner, the proposed non-local module can be flexibly integrated into existing deep networks for end-to-end training to capture deep feature correlation between each location and its neighborhood. (2) We fully employ the RNN structure for its parameter efficiency and allow deep feature correlation to be propagated along adjacent recurrent states. This new design boosts robustness against inaccurate correlation estimation due to severely degraded images. (3) We show that it is essential to maintain a confined neighborhood for computing deep feature correlation given degraded images. This is in contrast to existing practice [41] that deploys the whole image. Extensive experiments on both image denoising and super-resolution tasks are conducted. Thanks to the recurrent non-local operations and correlation propagation, the proposed NLRN achieves superior results to state-of-the-art methods with many fewer parameters. The code is available at https://github.com/Ding-Liu/NLRN.",
author = "Ding Liu and Bihan Wen and Yuchen Fan and Loy, {Chen Change} and Huang, {Thomas S}",
year = "2018",
month = "1",
day = "1",
language = "English (US)",
volume = "2018-December",
pages = "1673--1682",
journal = "Advances in Neural Information Processing Systems",
issn = "1049-5258",

}

TY - JOUR

T1 - Non-local recurrent network for image restoration

AU - Liu, Ding

AU - Wen, Bihan

AU - Fan, Yuchen

AU - Loy, Chen Change

AU - Huang, Thomas S

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Many classic methods have shown non-local self-similarity in natural images to be an effective prior for image restoration. However, it remains unclear and challenging to make use of this intrinsic property via deep networks. In this paper, we propose a non-local recurrent network (NLRN) as the first attempt to incorporate non-local operations into a recurrent neural network (RNN) for image restoration. The main contributions of this work are: (1) Unlike existing methods that measure self-similarity in an isolated manner, the proposed non-local module can be flexibly integrated into existing deep networks for end-to-end training to capture deep feature correlation between each location and its neighborhood. (2) We fully employ the RNN structure for its parameter efficiency and allow deep feature correlation to be propagated along adjacent recurrent states. This new design boosts robustness against inaccurate correlation estimation due to severely degraded images. (3) We show that it is essential to maintain a confined neighborhood for computing deep feature correlation given degraded images. This is in contrast to existing practice [41] that deploys the whole image. Extensive experiments on both image denoising and super-resolution tasks are conducted. Thanks to the recurrent non-local operations and correlation propagation, the proposed NLRN achieves superior results to state-of-the-art methods with many fewer parameters. The code is available at https://github.com/Ding-Liu/NLRN.

AB - Many classic methods have shown non-local self-similarity in natural images to be an effective prior for image restoration. However, it remains unclear and challenging to make use of this intrinsic property via deep networks. In this paper, we propose a non-local recurrent network (NLRN) as the first attempt to incorporate non-local operations into a recurrent neural network (RNN) for image restoration. The main contributions of this work are: (1) Unlike existing methods that measure self-similarity in an isolated manner, the proposed non-local module can be flexibly integrated into existing deep networks for end-to-end training to capture deep feature correlation between each location and its neighborhood. (2) We fully employ the RNN structure for its parameter efficiency and allow deep feature correlation to be propagated along adjacent recurrent states. This new design boosts robustness against inaccurate correlation estimation due to severely degraded images. (3) We show that it is essential to maintain a confined neighborhood for computing deep feature correlation given degraded images. This is in contrast to existing practice [41] that deploys the whole image. Extensive experiments on both image denoising and super-resolution tasks are conducted. Thanks to the recurrent non-local operations and correlation propagation, the proposed NLRN achieves superior results to state-of-the-art methods with many fewer parameters. The code is available at https://github.com/Ding-Liu/NLRN.

UR - http://www.scopus.com/inward/record.url?scp=85064803097&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064803097&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:85064803097

VL - 2018-December

SP - 1673

EP - 1682

JO - Advances in Neural Information Processing Systems

JF - Advances in Neural Information Processing Systems

SN - 1049-5258

ER -