Convex denoising using non-convex tight frame regularization

Ankit Parekh, Ivan Selesnick

Research output: Contribution to journalArticle

Abstract

This letter considers the problem of signal denoising using a sparse tight-frame analysis prior. The l<inf>1</inf> norm has been extensively used as a regularizer to promote sparsity; however, it tends to under-estimate non-zero values of the underlying signal. To more accurately estimate non-zero values, we propose the use of a non-convex regularizer, chosen so as to ensure convexity of the objective function. The convexity of the objective function is ensured by constraining the parameter of the non-convex penalty. We use ADMM to obtain a solution and show how to guarantee that ADMM converges to the global optimum of the objective function. We illustrate the proposed method for 1D and 2D signal denoising.

Original languageEnglish (US)
Article number7105866
Pages (from-to)1786-1790
Number of pages5
JournalIEEE Signal Processing Letters
Volume22
Issue number10
DOIs
StatePublished - Oct 1 2015

Fingerprint

Tight Frame
Denoising
Signal denoising
Regularization
Objective function
Convexity
L1-norm
Global Optimum
Sparsity
Estimate
Penalty
Tend
Converge

Keywords

  • Analysis model
  • convex optimization
  • non-convex regularization
  • sparse signal
  • tight frame

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing
  • Applied Mathematics

Cite this

Convex denoising using non-convex tight frame regularization. / Parekh, Ankit; Selesnick, Ivan.

In: IEEE Signal Processing Letters, Vol. 22, No. 10, 7105866, 01.10.2015, p. 1786-1790.

Research output: Contribution to journalArticle

@article{880fea2ab2ae45cfac5f9de02277dc71,
title = "Convex denoising using non-convex tight frame regularization",
abstract = "This letter considers the problem of signal denoising using a sparse tight-frame analysis prior. The l1 norm has been extensively used as a regularizer to promote sparsity; however, it tends to under-estimate non-zero values of the underlying signal. To more accurately estimate non-zero values, we propose the use of a non-convex regularizer, chosen so as to ensure convexity of the objective function. The convexity of the objective function is ensured by constraining the parameter of the non-convex penalty. We use ADMM to obtain a solution and show how to guarantee that ADMM converges to the global optimum of the objective function. We illustrate the proposed method for 1D and 2D signal denoising.",
keywords = "Analysis model, convex optimization, non-convex regularization, sparse signal, tight frame",
author = "Ankit Parekh and Ivan Selesnick",
year = "2015",
month = "10",
day = "1",
doi = "10.1109/LSP.2015.2432095",
language = "English (US)",
volume = "22",
pages = "1786--1790",
journal = "IEEE Signal Processing Letters",
issn = "1070-9908",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "10",

}

TY - JOUR

T1 - Convex denoising using non-convex tight frame regularization

AU - Parekh, Ankit

AU - Selesnick, Ivan

PY - 2015/10/1

Y1 - 2015/10/1

N2 - This letter considers the problem of signal denoising using a sparse tight-frame analysis prior. The l1 norm has been extensively used as a regularizer to promote sparsity; however, it tends to under-estimate non-zero values of the underlying signal. To more accurately estimate non-zero values, we propose the use of a non-convex regularizer, chosen so as to ensure convexity of the objective function. The convexity of the objective function is ensured by constraining the parameter of the non-convex penalty. We use ADMM to obtain a solution and show how to guarantee that ADMM converges to the global optimum of the objective function. We illustrate the proposed method for 1D and 2D signal denoising.

AB - This letter considers the problem of signal denoising using a sparse tight-frame analysis prior. The l1 norm has been extensively used as a regularizer to promote sparsity; however, it tends to under-estimate non-zero values of the underlying signal. To more accurately estimate non-zero values, we propose the use of a non-convex regularizer, chosen so as to ensure convexity of the objective function. The convexity of the objective function is ensured by constraining the parameter of the non-convex penalty. We use ADMM to obtain a solution and show how to guarantee that ADMM converges to the global optimum of the objective function. We illustrate the proposed method for 1D and 2D signal denoising.

KW - Analysis model

KW - convex optimization

KW - non-convex regularization

KW - sparse signal

KW - tight frame

UR - http://www.scopus.com/inward/record.url?scp=84930621911&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84930621911&partnerID=8YFLogxK

U2 - 10.1109/LSP.2015.2432095

DO - 10.1109/LSP.2015.2432095

M3 - Article

VL - 22

SP - 1786

EP - 1790

JO - IEEE Signal Processing Letters

JF - IEEE Signal Processing Letters

SN - 1070-9908

IS - 10

M1 - 7105866

ER -