Progressive blind deconvolution

Rana Hanocka, Nahum Kiryati

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We present a novel progressive framework for blind image restoration. Common blind restoration schemes first estimate the blur kernel, then employ non-blind deblurring. However, despite recent progress, the accuracy of PSF estimation is limited. Furthermore, the outcome of non-blind deblurring is highly sensitive to errors in the assumed PSF. Therefore, high quality blind deblurring has remained a major challenge. In this work, we combine state of the art regularizers for the image and the PSF, namely the Mumford & Shah piecewisesmooth image model and the sparse PSF prior. Previous works that used Mumford & Shah image regularization were either limited to nonblind deblurring or semi-blind deblurring assuming a parametric kernel known up to an unknown parameter. We suggest an iterative progressive restoration scheme, in which the imperfectly deblurred output of the current iteration is fed back as input to the next iteration. The kernel representing the residual blur is then estimated, and used to drive the non-blind restoration component, leading to finer deblurring. Experimental results demonstrate rapid convergence, and excellent performance on a wide variety of blurred images.

Original languageEnglish
Title of host publicationComputer Analysis of Images and Patterns - 16th International Conference, CAIP 2015, Proceedings
EditorsGeorge Azzopardi, Nicolai Petkov
PublisherSpringer Verlag
Pages313-325
Number of pages13
ISBN (Print)9783319231167
DOIs
StatePublished - 2015
Event16th International Conference on Computer Analysis of Images and Patterns, CAIP 2015 - Valletta, Malta
Duration: 2 Sep 20154 Sep 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9257
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference16th International Conference on Computer Analysis of Images and Patterns, CAIP 2015
Country/TerritoryMalta
CityValletta
Period2/09/154/09/15

Keywords

  • Image deblurring
  • Mumford & Shah regularization
  • Piecewise-smooth image model
  • Progressive blind restoration
  • Residual blur removal
  • Sparse PSF prior

Fingerprint

Dive into the research topics of 'Progressive blind deconvolution'. Together they form a unique fingerprint.

Cite this