Statistically and perceptually motivated nonlinear image representation

Siwei Lyu, Eero Simoncelli

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

Abstract

We describe an invertible nonlinear image transformation that is well-matched to the statistical properties of photographic images, as well as the perceptual sensitivity of the human visual system. Images are first decomposed using a multi-scale oriented linear transformation. In this domain, we develop a Markov random field model based on the dependencies within local clusters of transform coefficients associated with basis functions at nearby positions, orientations and scales. In this model, division of each coefficient by a particular linear combination of the amplitudes of others in the cluster produces a new nonlinear representation with marginally Gaussian statistics. We develop a reliable and efficient iterative procedure for inverting the divisive transformation. Finally, we probe the statistical and perceptual advantages of this image representation, examining robustness to added noise, rate-distortion behavior, and artifact-free local contrast enhancement.

Original languageEnglish (US)
Title of host publicationProceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII
Volume6492
DOIs
StatePublished - 2007
EventHuman Vision and Electronic Imaging XII - San Jose, CA, United States
Duration: Jan 29 2007Feb 1 2007

Other

OtherHuman Vision and Electronic Imaging XII
CountryUnited States
CitySan Jose, CA
Period1/29/072/1/07

Fingerprint

Linear transformations
Statistics
linear transformations
coefficients
division
artifacts
statistics
augmentation
probes
sensitivity

Keywords

  • Contrast enhancement
  • Divisive normalization
  • Independent components
  • Markov random field
  • Optimal representation

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Condensed Matter Physics

Cite this

Lyu, S., & Simoncelli, E. (2007). Statistically and perceptually motivated nonlinear image representation. In Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII (Vol. 6492). [649207] https://doi.org/10.1117/12.720848

Statistically and perceptually motivated nonlinear image representation. / Lyu, Siwei; Simoncelli, Eero.

Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII. Vol. 6492 2007. 649207.

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

Lyu, S & Simoncelli, E 2007, Statistically and perceptually motivated nonlinear image representation. in Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII. vol. 6492, 649207, Human Vision and Electronic Imaging XII, San Jose, CA, United States, 1/29/07. https://doi.org/10.1117/12.720848
Lyu S, Simoncelli E. Statistically and perceptually motivated nonlinear image representation. In Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII. Vol. 6492. 2007. 649207 https://doi.org/10.1117/12.720848
Lyu, Siwei ; Simoncelli, Eero. / Statistically and perceptually motivated nonlinear image representation. Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII. Vol. 6492 2007.
@inproceedings{31af77f28b5b47f3a312579c2d5feadb,
title = "Statistically and perceptually motivated nonlinear image representation",
abstract = "We describe an invertible nonlinear image transformation that is well-matched to the statistical properties of photographic images, as well as the perceptual sensitivity of the human visual system. Images are first decomposed using a multi-scale oriented linear transformation. In this domain, we develop a Markov random field model based on the dependencies within local clusters of transform coefficients associated with basis functions at nearby positions, orientations and scales. In this model, division of each coefficient by a particular linear combination of the amplitudes of others in the cluster produces a new nonlinear representation with marginally Gaussian statistics. We develop a reliable and efficient iterative procedure for inverting the divisive transformation. Finally, we probe the statistical and perceptual advantages of this image representation, examining robustness to added noise, rate-distortion behavior, and artifact-free local contrast enhancement.",
keywords = "Contrast enhancement, Divisive normalization, Independent components, Markov random field, Optimal representation",
author = "Siwei Lyu and Eero Simoncelli",
year = "2007",
doi = "10.1117/12.720848",
language = "English (US)",
isbn = "0819466050",
volume = "6492",
booktitle = "Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII",

}

TY - GEN

T1 - Statistically and perceptually motivated nonlinear image representation

AU - Lyu, Siwei

AU - Simoncelli, Eero

PY - 2007

Y1 - 2007

N2 - We describe an invertible nonlinear image transformation that is well-matched to the statistical properties of photographic images, as well as the perceptual sensitivity of the human visual system. Images are first decomposed using a multi-scale oriented linear transformation. In this domain, we develop a Markov random field model based on the dependencies within local clusters of transform coefficients associated with basis functions at nearby positions, orientations and scales. In this model, division of each coefficient by a particular linear combination of the amplitudes of others in the cluster produces a new nonlinear representation with marginally Gaussian statistics. We develop a reliable and efficient iterative procedure for inverting the divisive transformation. Finally, we probe the statistical and perceptual advantages of this image representation, examining robustness to added noise, rate-distortion behavior, and artifact-free local contrast enhancement.

AB - We describe an invertible nonlinear image transformation that is well-matched to the statistical properties of photographic images, as well as the perceptual sensitivity of the human visual system. Images are first decomposed using a multi-scale oriented linear transformation. In this domain, we develop a Markov random field model based on the dependencies within local clusters of transform coefficients associated with basis functions at nearby positions, orientations and scales. In this model, division of each coefficient by a particular linear combination of the amplitudes of others in the cluster produces a new nonlinear representation with marginally Gaussian statistics. We develop a reliable and efficient iterative procedure for inverting the divisive transformation. Finally, we probe the statistical and perceptual advantages of this image representation, examining robustness to added noise, rate-distortion behavior, and artifact-free local contrast enhancement.

KW - Contrast enhancement

KW - Divisive normalization

KW - Independent components

KW - Markov random field

KW - Optimal representation

UR - http://www.scopus.com/inward/record.url?scp=34548208599&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34548208599&partnerID=8YFLogxK

U2 - 10.1117/12.720848

DO - 10.1117/12.720848

M3 - Conference contribution

SN - 0819466050

SN - 9780819466051

VL - 6492

BT - Proceedings of SPIE-IS and T Electronic Imaging - Human Vision and Electronic Imaging XII

ER -