Learning Data Terms for Non-blind Deblurring

Jiangxin Dong, Jinshan Pan, Deqing Sun, Zhixun Su, Ming Hsuan Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Existing deblurring methods mainly focus on developing effective image priors and assume that blurred images contain insignificant amounts of noise. However, state-of-the-art deblurring methods do not perform well on real-world images degraded with significant noise or outliers. To address these issues, we show that it is critical to learn data fitting terms beyond the commonly used ℓ1 or ℓ2 norm. We propose a simple and effective discriminative framework to learn data terms that can adaptively handle blurred images in the presence of severe noise and outliers. Instead of learning the distribution of the data fitting errors, we directly learn the associated shrinkage function for the data term using a cascaded architecture, which is more flexible and efficient. Our analysis shows that the shrinkage functions learned at the intermediate stages can effectively suppress noise and preserve image structures. Extensive experimental results show that the proposed algorithm performs favorably against state-of-the-art methods.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
EditorsVittorio Ferrari, Cristian Sminchisescu, Yair Weiss, Martial Hebert
PublisherSpringer Verlag
Pages777-792
Number of pages16
ISBN (Print)9783030012519
DOIs
Publication statusPublished - 2018 Jan 1
Event15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany
Duration: 2018 Sep 82018 Sep 14

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11215 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other15th European Conference on Computer Vision, ECCV 2018
CountryGermany
CityMunich
Period18/9/818/9/14

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Learning Data Terms for Non-blind Deblurring'. Together they form a unique fingerprint.

  • Cite this

    Dong, J., Pan, J., Sun, D., Su, Z., & Yang, M. H. (2018). Learning Data Terms for Non-blind Deblurring. In V. Ferrari, C. Sminchisescu, Y. Weiss, & M. Hebert (Eds.), Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings (pp. 777-792). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11215 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-01252-6_46