Globally optimal smoothing functional for edge-enhancing regularized image restoration

Moon Gi Kang, Aggelos K. Katsaggelos, Kyu T. Park

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In regularized image restoration a solution is sought which preserves the fidelity to the noisy and blurred image data and also satisfies some constraints which represent our prior knowledge about the original image. A standard expression of this prior knowledge is that the original image is smooth. The regularization parameter balances these two requirements, i.e., fidelity to the data and smoothness of the solution. The smoothness requirement on the solution, however, results in a globally smooth image, i.e., no attention is paid to the preservation of the high spatial frequency information (edges). One approach towards the solution of this problem is the introduction of spatial adaptivity. A different approach is presented in this paper. According to this approach besides the constraint which bounds from above the energy of the restored image at high frequencies, a second constraint is used. With this constraint the high frequency energy of the restored image is also bounded from below. This means that very smooth solutions are not allowed, thus preserving edges and fine details in the restored image. Extending our previous work, we propose a nonlinear formulation of the regularization functional and derive an iterative algorithm for obtaining the unique minimum of this functional. The regularization parameters are evaluated simultaneously with the restored image, in an iterative fashion based on the partially restored image.

Original languageEnglish
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
Pages1505-1513
Number of pages9
Volume2727
Edition3/-
Publication statusPublished - 1996 Dec 1
EventVisual Communications and Image Processing'96. Part 2 (of 3) - Orlando, FL, USA
Duration: 1996 Mar 171996 Mar 20

Other

OtherVisual Communications and Image Processing'96. Part 2 (of 3)
CityOrlando, FL, USA
Period96/3/1796/3/20

Fingerprint

Image Restoration
Image reconstruction
smoothing
restoration
Smoothing
Regularization Parameter
Prior Knowledge
Fidelity
Smoothness
Bound Constraints
requirements
Adaptivity
Requirements
Smooth Solution
Energy
Preservation
Iterative Algorithm
Regularization
preserving
formulations

All Science Journal Classification (ASJC) codes

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Kang, M. G., Katsaggelos, A. K., & Park, K. T. (1996). Globally optimal smoothing functional for edge-enhancing regularized image restoration. In Proceedings of SPIE - The International Society for Optical Engineering (3/- ed., Vol. 2727 , pp. 1505-1513)
Kang, Moon Gi ; Katsaggelos, Aggelos K. ; Park, Kyu T. / Globally optimal smoothing functional for edge-enhancing regularized image restoration. Proceedings of SPIE - The International Society for Optical Engineering. Vol. 2727 3/-. ed. 1996. pp. 1505-1513
@inproceedings{788e326eb5c048bf9823ae54178be894,
title = "Globally optimal smoothing functional for edge-enhancing regularized image restoration",
abstract = "In regularized image restoration a solution is sought which preserves the fidelity to the noisy and blurred image data and also satisfies some constraints which represent our prior knowledge about the original image. A standard expression of this prior knowledge is that the original image is smooth. The regularization parameter balances these two requirements, i.e., fidelity to the data and smoothness of the solution. The smoothness requirement on the solution, however, results in a globally smooth image, i.e., no attention is paid to the preservation of the high spatial frequency information (edges). One approach towards the solution of this problem is the introduction of spatial adaptivity. A different approach is presented in this paper. According to this approach besides the constraint which bounds from above the energy of the restored image at high frequencies, a second constraint is used. With this constraint the high frequency energy of the restored image is also bounded from below. This means that very smooth solutions are not allowed, thus preserving edges and fine details in the restored image. Extending our previous work, we propose a nonlinear formulation of the regularization functional and derive an iterative algorithm for obtaining the unique minimum of this functional. The regularization parameters are evaluated simultaneously with the restored image, in an iterative fashion based on the partially restored image.",
author = "Kang, {Moon Gi} and Katsaggelos, {Aggelos K.} and Park, {Kyu T.}",
year = "1996",
month = "12",
day = "1",
language = "English",
isbn = "0819421030",
volume = "2727",
pages = "1505--1513",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",
edition = "3/-",

}

Kang, MG, Katsaggelos, AK & Park, KT 1996, Globally optimal smoothing functional for edge-enhancing regularized image restoration. in Proceedings of SPIE - The International Society for Optical Engineering. 3/- edn, vol. 2727 , pp. 1505-1513, Visual Communications and Image Processing'96. Part 2 (of 3), Orlando, FL, USA, 96/3/17.

Globally optimal smoothing functional for edge-enhancing regularized image restoration. / Kang, Moon Gi; Katsaggelos, Aggelos K.; Park, Kyu T.

Proceedings of SPIE - The International Society for Optical Engineering. Vol. 2727 3/-. ed. 1996. p. 1505-1513.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Globally optimal smoothing functional for edge-enhancing regularized image restoration

AU - Kang, Moon Gi

AU - Katsaggelos, Aggelos K.

AU - Park, Kyu T.

PY - 1996/12/1

Y1 - 1996/12/1

N2 - In regularized image restoration a solution is sought which preserves the fidelity to the noisy and blurred image data and also satisfies some constraints which represent our prior knowledge about the original image. A standard expression of this prior knowledge is that the original image is smooth. The regularization parameter balances these two requirements, i.e., fidelity to the data and smoothness of the solution. The smoothness requirement on the solution, however, results in a globally smooth image, i.e., no attention is paid to the preservation of the high spatial frequency information (edges). One approach towards the solution of this problem is the introduction of spatial adaptivity. A different approach is presented in this paper. According to this approach besides the constraint which bounds from above the energy of the restored image at high frequencies, a second constraint is used. With this constraint the high frequency energy of the restored image is also bounded from below. This means that very smooth solutions are not allowed, thus preserving edges and fine details in the restored image. Extending our previous work, we propose a nonlinear formulation of the regularization functional and derive an iterative algorithm for obtaining the unique minimum of this functional. The regularization parameters are evaluated simultaneously with the restored image, in an iterative fashion based on the partially restored image.

AB - In regularized image restoration a solution is sought which preserves the fidelity to the noisy and blurred image data and also satisfies some constraints which represent our prior knowledge about the original image. A standard expression of this prior knowledge is that the original image is smooth. The regularization parameter balances these two requirements, i.e., fidelity to the data and smoothness of the solution. The smoothness requirement on the solution, however, results in a globally smooth image, i.e., no attention is paid to the preservation of the high spatial frequency information (edges). One approach towards the solution of this problem is the introduction of spatial adaptivity. A different approach is presented in this paper. According to this approach besides the constraint which bounds from above the energy of the restored image at high frequencies, a second constraint is used. With this constraint the high frequency energy of the restored image is also bounded from below. This means that very smooth solutions are not allowed, thus preserving edges and fine details in the restored image. Extending our previous work, we propose a nonlinear formulation of the regularization functional and derive an iterative algorithm for obtaining the unique minimum of this functional. The regularization parameters are evaluated simultaneously with the restored image, in an iterative fashion based on the partially restored image.

UR - http://www.scopus.com/inward/record.url?scp=0030405052&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030405052&partnerID=8YFLogxK

M3 - Conference contribution

SN - 0819421030

SN - 9780819421036

VL - 2727

SP - 1505

EP - 1513

BT - Proceedings of SPIE - The International Society for Optical Engineering

ER -

Kang MG, Katsaggelos AK, Park KT. Globally optimal smoothing functional for edge-enhancing regularized image restoration. In Proceedings of SPIE - The International Society for Optical Engineering. 3/- ed. Vol. 2727 . 1996. p. 1505-1513