Image modeling and denoising with orientation-adapted Gaussian scale mixtures

David K. Hammond, Eero Simoncelli

Research output: Contribution to journalArticle

Abstract

We develop a statistical model to describe the spatially varying behavior of local neighborhoods of coefficients in a multiscale image representation. Neighborhoods are modeled as samples of a multivariate Gaussian density that are modulated and rotated according to the values of two hidden random variables, thus allowing the model to adapt to the local amplitude and orientation of the signal. A third hidden variable selects between this oriented process and a nonoriented scale mixture of Gaussians process, thus providing adaptability to the local orientedness of the signal. Based on this model, we develop an optimal Bayesian least squares estimator for denoising images and show through simulations that the resulting method exhibits significant improvement over previously published results obtained with Gaussian scale mixtures.

Original languageEnglish (US)
Pages (from-to)2089-2101
Number of pages13
JournalIEEE Transactions on Image Processing
Volume17
Issue number11
DOIs
StatePublished - 2008

Fingerprint

Image denoising
Random variables
Statistical Models

Keywords

  • Gaussian Scale Mixtures
  • Image denoising
  • Image processing
  • Statistical image modeling
  • Wavelet transforms

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Software

Cite this

Image modeling and denoising with orientation-adapted Gaussian scale mixtures. / Hammond, David K.; Simoncelli, Eero.

In: IEEE Transactions on Image Processing, Vol. 17, No. 11, 2008, p. 2089-2101.

Research output: Contribution to journalArticle

@article{eaed2c84541f42e2a34f08e32c8f451f,
title = "Image modeling and denoising with orientation-adapted Gaussian scale mixtures",
abstract = "We develop a statistical model to describe the spatially varying behavior of local neighborhoods of coefficients in a multiscale image representation. Neighborhoods are modeled as samples of a multivariate Gaussian density that are modulated and rotated according to the values of two hidden random variables, thus allowing the model to adapt to the local amplitude and orientation of the signal. A third hidden variable selects between this oriented process and a nonoriented scale mixture of Gaussians process, thus providing adaptability to the local orientedness of the signal. Based on this model, we develop an optimal Bayesian least squares estimator for denoising images and show through simulations that the resulting method exhibits significant improvement over previously published results obtained with Gaussian scale mixtures.",
keywords = "Gaussian Scale Mixtures, Image denoising, Image processing, Statistical image modeling, Wavelet transforms",
author = "Hammond, {David K.} and Eero Simoncelli",
year = "2008",
doi = "10.1109/TIP.2008.2004796",
language = "English (US)",
volume = "17",
pages = "2089--2101",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "11",

}

TY - JOUR

T1 - Image modeling and denoising with orientation-adapted Gaussian scale mixtures

AU - Hammond, David K.

AU - Simoncelli, Eero

PY - 2008

Y1 - 2008

N2 - We develop a statistical model to describe the spatially varying behavior of local neighborhoods of coefficients in a multiscale image representation. Neighborhoods are modeled as samples of a multivariate Gaussian density that are modulated and rotated according to the values of two hidden random variables, thus allowing the model to adapt to the local amplitude and orientation of the signal. A third hidden variable selects between this oriented process and a nonoriented scale mixture of Gaussians process, thus providing adaptability to the local orientedness of the signal. Based on this model, we develop an optimal Bayesian least squares estimator for denoising images and show through simulations that the resulting method exhibits significant improvement over previously published results obtained with Gaussian scale mixtures.

AB - We develop a statistical model to describe the spatially varying behavior of local neighborhoods of coefficients in a multiscale image representation. Neighborhoods are modeled as samples of a multivariate Gaussian density that are modulated and rotated according to the values of two hidden random variables, thus allowing the model to adapt to the local amplitude and orientation of the signal. A third hidden variable selects between this oriented process and a nonoriented scale mixture of Gaussians process, thus providing adaptability to the local orientedness of the signal. Based on this model, we develop an optimal Bayesian least squares estimator for denoising images and show through simulations that the resulting method exhibits significant improvement over previously published results obtained with Gaussian scale mixtures.

KW - Gaussian Scale Mixtures

KW - Image denoising

KW - Image processing

KW - Statistical image modeling

KW - Wavelet transforms

UR - http://www.scopus.com/inward/record.url?scp=54949106570&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=54949106570&partnerID=8YFLogxK

U2 - 10.1109/TIP.2008.2004796

DO - 10.1109/TIP.2008.2004796

M3 - Article

VL - 17

SP - 2089

EP - 2101

JO - IEEE Transactions on Image Processing

JF - IEEE Transactions on Image Processing

SN - 1057-7149

IS - 11

ER -