Adaptive kernel-based image denoising employing semi-parametric regularization

Pantelis Bouboulis, Konstantinos Slavakis, Sergios Theodoridis

Research output: Contribution to journalArticle

42 Citations (Scopus)

Abstract

The main contribution of this paper is the development of a novel approach, based on the theory of Reproducing Kernel Hilbert Spaces (RKHS), for the problem of noise removal in the spatial domain. The proposed methodology has the advantage that it is able to remove any kind of additive noise (impulse, gaussian, uniform, etc.) from any digital image, in contrast to the most commonly used denoising techniques, which are noise dependent. The problem is cast as an optimization task in a RKHS, by taking advantage of the celebrated Representer Theorem in its semi-parametric formulation. The semi-parametric formulation, although known in theory, has so far found limited, to our knowledge, application. However, in the image denoising problem, its use is dictated by the nature of the problem itself. The need for edge preservation naturally leads to such a modeling. Examples verify that in the presence of gaussian noise the proposed methodology performs well compared to wavelet based technics and outperforms them significantly in the presence of impulse or mixed noise.

Original languageEnglish (US)
Article number5430976
Pages (from-to)1465-1479
Number of pages15
JournalIEEE Transactions on Image Processing
Volume19
Issue number6
DOIs
StatePublished - Jun 1 2010

Fingerprint

Image denoising
Hilbert spaces
Additive noise

Keywords

  • Denoising
  • Kernel
  • Reproducing Kernel Hilbert Spaces (RKHS)
  • Semi-parametric representer theorem

Cite this

Adaptive kernel-based image denoising employing semi-parametric regularization. / Bouboulis, Pantelis; Slavakis, Konstantinos; Theodoridis, Sergios.

In: IEEE Transactions on Image Processing, Vol. 19, No. 6, 5430976, 01.06.2010, p. 1465-1479.

Research output: Contribution to journalArticle

Bouboulis, Pantelis ; Slavakis, Konstantinos ; Theodoridis, Sergios. / Adaptive kernel-based image denoising employing semi-parametric regularization. In: IEEE Transactions on Image Processing. 2010 ; Vol. 19, No. 6. pp. 1465-1479.
@article{5e9242303ce644679ef98842e18a41ae,
title = "Adaptive kernel-based image denoising employing semi-parametric regularization",
abstract = "The main contribution of this paper is the development of a novel approach, based on the theory of Reproducing Kernel Hilbert Spaces (RKHS), for the problem of noise removal in the spatial domain. The proposed methodology has the advantage that it is able to remove any kind of additive noise (impulse, gaussian, uniform, etc.) from any digital image, in contrast to the most commonly used denoising techniques, which are noise dependent. The problem is cast as an optimization task in a RKHS, by taking advantage of the celebrated Representer Theorem in its semi-parametric formulation. The semi-parametric formulation, although known in theory, has so far found limited, to our knowledge, application. However, in the image denoising problem, its use is dictated by the nature of the problem itself. The need for edge preservation naturally leads to such a modeling. Examples verify that in the presence of gaussian noise the proposed methodology performs well compared to wavelet based technics and outperforms them significantly in the presence of impulse or mixed noise.",
keywords = "Denoising, Kernel, Reproducing Kernel Hilbert Spaces (RKHS), Semi-parametric representer theorem",
author = "Pantelis Bouboulis and Konstantinos Slavakis and Sergios Theodoridis",
year = "2010",
month = "6",
day = "1",
doi = "10.1109/TIP.2010.2042995",
language = "English (US)",
volume = "19",
pages = "1465--1479",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",

}

TY - JOUR

T1 - Adaptive kernel-based image denoising employing semi-parametric regularization

AU - Bouboulis, Pantelis

AU - Slavakis, Konstantinos

AU - Theodoridis, Sergios

PY - 2010/6/1

Y1 - 2010/6/1

N2 - The main contribution of this paper is the development of a novel approach, based on the theory of Reproducing Kernel Hilbert Spaces (RKHS), for the problem of noise removal in the spatial domain. The proposed methodology has the advantage that it is able to remove any kind of additive noise (impulse, gaussian, uniform, etc.) from any digital image, in contrast to the most commonly used denoising techniques, which are noise dependent. The problem is cast as an optimization task in a RKHS, by taking advantage of the celebrated Representer Theorem in its semi-parametric formulation. The semi-parametric formulation, although known in theory, has so far found limited, to our knowledge, application. However, in the image denoising problem, its use is dictated by the nature of the problem itself. The need for edge preservation naturally leads to such a modeling. Examples verify that in the presence of gaussian noise the proposed methodology performs well compared to wavelet based technics and outperforms them significantly in the presence of impulse or mixed noise.

AB - The main contribution of this paper is the development of a novel approach, based on the theory of Reproducing Kernel Hilbert Spaces (RKHS), for the problem of noise removal in the spatial domain. The proposed methodology has the advantage that it is able to remove any kind of additive noise (impulse, gaussian, uniform, etc.) from any digital image, in contrast to the most commonly used denoising techniques, which are noise dependent. The problem is cast as an optimization task in a RKHS, by taking advantage of the celebrated Representer Theorem in its semi-parametric formulation. The semi-parametric formulation, although known in theory, has so far found limited, to our knowledge, application. However, in the image denoising problem, its use is dictated by the nature of the problem itself. The need for edge preservation naturally leads to such a modeling. Examples verify that in the presence of gaussian noise the proposed methodology performs well compared to wavelet based technics and outperforms them significantly in the presence of impulse or mixed noise.

KW - Denoising

KW - Kernel

KW - Reproducing Kernel Hilbert Spaces (RKHS)

KW - Semi-parametric representer theorem

UR - http://www.scopus.com/inward/record.url?scp=77952590168&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77952590168&partnerID=8YFLogxK

U2 - 10.1109/TIP.2010.2042995

DO - 10.1109/TIP.2010.2042995

M3 - Article

VL - 19

SP - 1465

EP - 1479

JO - IEEE Transactions on Image Processing

JF - IEEE Transactions on Image Processing

SN - 1057-7149

IS - 6

M1 - 5430976

ER -