Sparsity constrained regularization for multiframe image restoration

Premchandra M. Shankar, Mark A Neifeld

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

In this paper we present a new algorithm for restoring an object from multiple undersampled low-resolution (LR) images that are degraded by optical blur and additive white Gaussian noise. We formulate the multiframe superresolution problem as maximum a posteriori estimation. The prior knowledge that the object is sparse in some domain is incorporated in two ways: first we use the popular l1 norm as the regularization operator. Second, we model wavelet coefficients of natural objects using generalized Gaussian densities. The model parameters are learned from a set of training objects, and the regularization operator is derived from these parameters. We compare the results from our algorithms with an expectation-maximization (EM) algorithm for l1 norm minimization and also with the linear minimum-mean-squared error (LMMSE) estimator. Using only eight 4 × 4 pixel downsampled LR images the reconstruction errors of object estimates obtained from our algorithm are 5.5% smaller than by the EM method and 14.3% smaller than by the LMMSE method.

Original languageEnglish (US)
Pages (from-to)1199-1214
Number of pages16
JournalJournal of the Optical Society of America A: Optics and Image Science, and Vision
Volume25
Issue number5
DOIs
StatePublished - May 1 2008

Fingerprint

Image reconstruction
restoration
image resolution
Image resolution
norms
operators
Computer-Assisted Image Processing
random noise
estimators
education
Pixels
pixels
optimization
coefficients
estimates

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Computer Vision and Pattern Recognition

Cite this

Sparsity constrained regularization for multiframe image restoration. / Shankar, Premchandra M.; Neifeld, Mark A.

In: Journal of the Optical Society of America A: Optics and Image Science, and Vision, Vol. 25, No. 5, 01.05.2008, p. 1199-1214.

Research output: Contribution to journalArticle

@article{1f0b575ba51a49bf8c2a3d30ca4ba414,
title = "Sparsity constrained regularization for multiframe image restoration",
abstract = "In this paper we present a new algorithm for restoring an object from multiple undersampled low-resolution (LR) images that are degraded by optical blur and additive white Gaussian noise. We formulate the multiframe superresolution problem as maximum a posteriori estimation. The prior knowledge that the object is sparse in some domain is incorporated in two ways: first we use the popular l1 norm as the regularization operator. Second, we model wavelet coefficients of natural objects using generalized Gaussian densities. The model parameters are learned from a set of training objects, and the regularization operator is derived from these parameters. We compare the results from our algorithms with an expectation-maximization (EM) algorithm for l1 norm minimization and also with the linear minimum-mean-squared error (LMMSE) estimator. Using only eight 4 × 4 pixel downsampled LR images the reconstruction errors of object estimates obtained from our algorithm are 5.5{\%} smaller than by the EM method and 14.3{\%} smaller than by the LMMSE method.",
author = "Shankar, {Premchandra M.} and Neifeld, {Mark A}",
year = "2008",
month = "5",
day = "1",
doi = "10.1364/JOSAA.25.001199",
language = "English (US)",
volume = "25",
pages = "1199--1214",
journal = "Journal of the Optical Society of America A: Optics and Image Science, and Vision",
issn = "1084-7529",
publisher = "The Optical Society",
number = "5",

}

TY - JOUR

T1 - Sparsity constrained regularization for multiframe image restoration

AU - Shankar, Premchandra M.

AU - Neifeld, Mark A

PY - 2008/5/1

Y1 - 2008/5/1

N2 - In this paper we present a new algorithm for restoring an object from multiple undersampled low-resolution (LR) images that are degraded by optical blur and additive white Gaussian noise. We formulate the multiframe superresolution problem as maximum a posteriori estimation. The prior knowledge that the object is sparse in some domain is incorporated in two ways: first we use the popular l1 norm as the regularization operator. Second, we model wavelet coefficients of natural objects using generalized Gaussian densities. The model parameters are learned from a set of training objects, and the regularization operator is derived from these parameters. We compare the results from our algorithms with an expectation-maximization (EM) algorithm for l1 norm minimization and also with the linear minimum-mean-squared error (LMMSE) estimator. Using only eight 4 × 4 pixel downsampled LR images the reconstruction errors of object estimates obtained from our algorithm are 5.5% smaller than by the EM method and 14.3% smaller than by the LMMSE method.

AB - In this paper we present a new algorithm for restoring an object from multiple undersampled low-resolution (LR) images that are degraded by optical blur and additive white Gaussian noise. We formulate the multiframe superresolution problem as maximum a posteriori estimation. The prior knowledge that the object is sparse in some domain is incorporated in two ways: first we use the popular l1 norm as the regularization operator. Second, we model wavelet coefficients of natural objects using generalized Gaussian densities. The model parameters are learned from a set of training objects, and the regularization operator is derived from these parameters. We compare the results from our algorithms with an expectation-maximization (EM) algorithm for l1 norm minimization and also with the linear minimum-mean-squared error (LMMSE) estimator. Using only eight 4 × 4 pixel downsampled LR images the reconstruction errors of object estimates obtained from our algorithm are 5.5% smaller than by the EM method and 14.3% smaller than by the LMMSE method.

UR - http://www.scopus.com/inward/record.url?scp=62249219141&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=62249219141&partnerID=8YFLogxK

U2 - 10.1364/JOSAA.25.001199

DO - 10.1364/JOSAA.25.001199

M3 - Article

C2 - 18451929

AN - SCOPUS:62249219141

VL - 25

SP - 1199

EP - 1214

JO - Journal of the Optical Society of America A: Optics and Image Science, and Vision

JF - Journal of the Optical Society of America A: Optics and Image Science, and Vision

SN - 1084-7529

IS - 5

ER -