On total variation denoising: A new majorization-minimization algorithm and an experimental comparison with wavalet denoising

M. A.T. Figueiredo, J. B. Dias, J. P. Oliveira, R. D. Nowak

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

113 Scopus citations

Abstract

Image denoising is a classical problem which has been addressed using a variety of conceptual frameworks and computational tools. Most approaches use some form of penalty/prior as a regularizer, expressing a preference for images with some form of (generalized) "smoothness". Total variation (TV) and wavelet-based methods have received a great deal of attention in the last decade and are among the state of the art in this problem. However, as far as we know, no experimental studies have been carried out, comparing the relative performance of the two classes of methods. In this paper, we present the results of such a comparison. Prior to that, we introduce a new majorization- minimization algorithm to implement the TV denoising criterion. We conclude that TV is outperformed by recent state of the art wavelet-based denoising methods, but performs competitively with older wavelet-based methods.

Original languageBritish English
Title of host publication2006 IEEE International Conference on Image Processing, ICIP 2006 - Proceedings
Pages2633-2636
Number of pages4
DOIs
StatePublished - 2006
Event2006 IEEE International Conference on Image Processing, ICIP 2006 - Atlanta, GA, United States
Duration: 8 Oct 200611 Oct 2006

Publication series

NameProceedings - International Conference on Image Processing, ICIP
ISSN (Print)1522-4880

Conference

Conference2006 IEEE International Conference on Image Processing, ICIP 2006
Country/TerritoryUnited States
CityAtlanta, GA
Period8/10/0611/10/06

Keywords

  • Image denoising
  • Image restoration
  • Majorization-minimization algorithms
  • Total variation

Fingerprint

Dive into the research topics of 'On total variation denoising: A new majorization-minimization algorithm and an experimental comparison with wavalet denoising'. Together they form a unique fingerprint.

Cite this