New complexity prior for multiresolution image denoising

Juan Liu, Pierre Moulin

Research output: Contribution to conferencePaper

Abstract

Application of the Minimum Description Length (MDL) principle to multiresolution image denoising has been somewhat unsuccessful to date. This disappointing performance is due to the crudeness of the underlying prior image models, which lead to overly sparse solutions. We propose a new family of complexity priors based on Rissanen's universal prior for integers, which produces estimates with better sparsity properties. This method vastly outperforms previous MDL schemes and is competitive with Bayesian estimators using Generalized Gaussian priors on wavelet coefficients.

Original languageEnglish (US)
Pages637-640
Number of pages4
StatePublished - Jan 1 1998
EventProceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis - Pittsburgh, PA, USA
Duration: Oct 6 1998Oct 9 1998

Other

OtherProceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis
CityPittsburgh, PA, USA
Period10/6/9810/9/98

Fingerprint

Image denoising

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Liu, J., & Moulin, P. (1998). New complexity prior for multiresolution image denoising. 637-640. Paper presented at Proceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis, Pittsburgh, PA, USA, .

New complexity prior for multiresolution image denoising. / Liu, Juan; Moulin, Pierre.

1998. 637-640 Paper presented at Proceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis, Pittsburgh, PA, USA, .

Research output: Contribution to conferencePaper

Liu, J & Moulin, P 1998, 'New complexity prior for multiresolution image denoising', Paper presented at Proceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis, Pittsburgh, PA, USA, 10/6/98 - 10/9/98 pp. 637-640.
Liu J, Moulin P. New complexity prior for multiresolution image denoising. 1998. Paper presented at Proceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis, Pittsburgh, PA, USA, .
Liu, Juan ; Moulin, Pierre. / New complexity prior for multiresolution image denoising. Paper presented at Proceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis, Pittsburgh, PA, USA, .4 p.
@conference{0bcac6ef2a7f472b850b2533bc226a80,
title = "New complexity prior for multiresolution image denoising",
abstract = "Application of the Minimum Description Length (MDL) principle to multiresolution image denoising has been somewhat unsuccessful to date. This disappointing performance is due to the crudeness of the underlying prior image models, which lead to overly sparse solutions. We propose a new family of complexity priors based on Rissanen's universal prior for integers, which produces estimates with better sparsity properties. This method vastly outperforms previous MDL schemes and is competitive with Bayesian estimators using Generalized Gaussian priors on wavelet coefficients.",
author = "Juan Liu and Pierre Moulin",
year = "1998",
month = "1",
day = "1",
language = "English (US)",
pages = "637--640",
note = "Proceedings of the 1998 IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis ; Conference date: 06-10-1998 Through 09-10-1998",

}

TY - CONF

T1 - New complexity prior for multiresolution image denoising

AU - Liu, Juan

AU - Moulin, Pierre

PY - 1998/1/1

Y1 - 1998/1/1

N2 - Application of the Minimum Description Length (MDL) principle to multiresolution image denoising has been somewhat unsuccessful to date. This disappointing performance is due to the crudeness of the underlying prior image models, which lead to overly sparse solutions. We propose a new family of complexity priors based on Rissanen's universal prior for integers, which produces estimates with better sparsity properties. This method vastly outperforms previous MDL schemes and is competitive with Bayesian estimators using Generalized Gaussian priors on wavelet coefficients.

AB - Application of the Minimum Description Length (MDL) principle to multiresolution image denoising has been somewhat unsuccessful to date. This disappointing performance is due to the crudeness of the underlying prior image models, which lead to overly sparse solutions. We propose a new family of complexity priors based on Rissanen's universal prior for integers, which produces estimates with better sparsity properties. This method vastly outperforms previous MDL schemes and is competitive with Bayesian estimators using Generalized Gaussian priors on wavelet coefficients.

UR - http://www.scopus.com/inward/record.url?scp=0031622177&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031622177&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0031622177

SP - 637

EP - 640

ER -