Inpainting and zooming using sparse representations

M. J. Fadili, J. L. Starck, Fionn Murtagh

Research output: Contribution to journalArticle

219 Citations (Scopus)

Abstract

Representing the image to be inpainted in an appropriate sparse representation dictionary, and combining elements from Bayesian statistics and modern harmonic analysis, we introduce an expectation maximization (EM) algorithm for image inpainting and interpolation. From a statistical point of view, the inpainting/interpolation can be viewed as an estimation problem with missing data. Toward this goal, we propose the idea of using the EM mechanism in a Bayesian framework, where a sparsity promoting prior penalty is imposed on the reconstructed coefficients. The EM framework gives a principled way to establish formally the idea that missing samples can be recovered/interpolated based on sparse representations. We first introduce an easy and efficient sparse-representation-based iterative algorithm for image inpainting. Additionally, we derive its theoretical convergence properties. Compared to its competitors, this algorithm allows a high degree of flexibility to recover different structural components in the image (piecewise smooth, curvilinear, texture, etc.). We also suggest some guidelines to automatically tune the regularization parameter.

LanguageEnglish
Pages64-79
Number of pages16
JournalComputer Journal
Volume52
Issue number1
DOIs
Publication statusPublished - 1 Jan 2009
Externally publishedYes

Fingerprint

Interpolation
Harmonic analysis
Glossaries
Textures
Statistics

Cite this

Fadili, M. J. ; Starck, J. L. ; Murtagh, Fionn. / Inpainting and zooming using sparse representations. In: Computer Journal. 2009 ; Vol. 52, No. 1. pp. 64-79.
@article{8d30bb5e8819497595a133bf2873caa8,
title = "Inpainting and zooming using sparse representations",
abstract = "Representing the image to be inpainted in an appropriate sparse representation dictionary, and combining elements from Bayesian statistics and modern harmonic analysis, we introduce an expectation maximization (EM) algorithm for image inpainting and interpolation. From a statistical point of view, the inpainting/interpolation can be viewed as an estimation problem with missing data. Toward this goal, we propose the idea of using the EM mechanism in a Bayesian framework, where a sparsity promoting prior penalty is imposed on the reconstructed coefficients. The EM framework gives a principled way to establish formally the idea that missing samples can be recovered/interpolated based on sparse representations. We first introduce an easy and efficient sparse-representation-based iterative algorithm for image inpainting. Additionally, we derive its theoretical convergence properties. Compared to its competitors, this algorithm allows a high degree of flexibility to recover different structural components in the image (piecewise smooth, curvilinear, texture, etc.). We also suggest some guidelines to automatically tune the regularization parameter.",
keywords = "EM algorithm, Inpainting, Interpolation, Penalized likelihood, Sparse representations",
author = "Fadili, {M. J.} and Starck, {J. L.} and Fionn Murtagh",
year = "2009",
month = "1",
day = "1",
doi = "10.1093/comjnl/bxm055",
language = "English",
volume = "52",
pages = "64--79",
journal = "Computer Journal",
issn = "0010-4620",
publisher = "Oxford University Press",
number = "1",

}

Inpainting and zooming using sparse representations. / Fadili, M. J.; Starck, J. L.; Murtagh, Fionn.

In: Computer Journal, Vol. 52, No. 1, 01.01.2009, p. 64-79.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Inpainting and zooming using sparse representations

AU - Fadili, M. J.

AU - Starck, J. L.

AU - Murtagh, Fionn

PY - 2009/1/1

Y1 - 2009/1/1

N2 - Representing the image to be inpainted in an appropriate sparse representation dictionary, and combining elements from Bayesian statistics and modern harmonic analysis, we introduce an expectation maximization (EM) algorithm for image inpainting and interpolation. From a statistical point of view, the inpainting/interpolation can be viewed as an estimation problem with missing data. Toward this goal, we propose the idea of using the EM mechanism in a Bayesian framework, where a sparsity promoting prior penalty is imposed on the reconstructed coefficients. The EM framework gives a principled way to establish formally the idea that missing samples can be recovered/interpolated based on sparse representations. We first introduce an easy and efficient sparse-representation-based iterative algorithm for image inpainting. Additionally, we derive its theoretical convergence properties. Compared to its competitors, this algorithm allows a high degree of flexibility to recover different structural components in the image (piecewise smooth, curvilinear, texture, etc.). We also suggest some guidelines to automatically tune the regularization parameter.

AB - Representing the image to be inpainted in an appropriate sparse representation dictionary, and combining elements from Bayesian statistics and modern harmonic analysis, we introduce an expectation maximization (EM) algorithm for image inpainting and interpolation. From a statistical point of view, the inpainting/interpolation can be viewed as an estimation problem with missing data. Toward this goal, we propose the idea of using the EM mechanism in a Bayesian framework, where a sparsity promoting prior penalty is imposed on the reconstructed coefficients. The EM framework gives a principled way to establish formally the idea that missing samples can be recovered/interpolated based on sparse representations. We first introduce an easy and efficient sparse-representation-based iterative algorithm for image inpainting. Additionally, we derive its theoretical convergence properties. Compared to its competitors, this algorithm allows a high degree of flexibility to recover different structural components in the image (piecewise smooth, curvilinear, texture, etc.). We also suggest some guidelines to automatically tune the regularization parameter.

KW - EM algorithm

KW - Inpainting

KW - Interpolation

KW - Penalized likelihood

KW - Sparse representations

UR - http://www.scopus.com/inward/record.url?scp=60649102899&partnerID=8YFLogxK

U2 - 10.1093/comjnl/bxm055

DO - 10.1093/comjnl/bxm055

M3 - Article

VL - 52

SP - 64

EP - 79

JO - Computer Journal

T2 - Computer Journal

JF - Computer Journal

SN - 0010-4620

IS - 1

ER -