Maximum likelihood difference scaling of image quality in compression-degraded images

Christophe Charrier, Laurence T. Maloney, Hocine Cherifi, Kenneth Knoblaunch

Research output: Contribution to journalArticle

Abstract

Lossy image compression techniques allow arbitrarily high compression rates but at the price of poor image quality. We applied maximum likelihood difference scaling to evaluate image quality of nine images, each compressed via vector quantization to ten different levels, within two different color spaces, RGB and CIE 1976 L*a*b*. In L*a*b* space, images could be compressed on average by 32% more than in RGB space, with little additional loss in quality. Further compression led to marked perceptual changes. Our approach permits a rapid, direct measurement of the consequences of image compression for human observers.

Original languageEnglish (US)
Pages (from-to)3418-3426
Number of pages9
JournalJournal of the Optical Society of America A: Optics and Image Science, and Vision
Volume24
Issue number11
DOIs
StatePublished - 2007

Fingerprint

Data Compression
Image compression
Image quality
Maximum likelihood
Vector quantization
Color

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Computer Vision and Pattern Recognition

Cite this

Maximum likelihood difference scaling of image quality in compression-degraded images. / Charrier, Christophe; Maloney, Laurence T.; Cherifi, Hocine; Knoblaunch, Kenneth.

In: Journal of the Optical Society of America A: Optics and Image Science, and Vision, Vol. 24, No. 11, 2007, p. 3418-3426.

Research output: Contribution to journalArticle

@article{ca2ebec1dfbd4f398d0e6691d9e806e7,
title = "Maximum likelihood difference scaling of image quality in compression-degraded images",
abstract = "Lossy image compression techniques allow arbitrarily high compression rates but at the price of poor image quality. We applied maximum likelihood difference scaling to evaluate image quality of nine images, each compressed via vector quantization to ten different levels, within two different color spaces, RGB and CIE 1976 L*a*b*. In L*a*b* space, images could be compressed on average by 32{\%} more than in RGB space, with little additional loss in quality. Further compression led to marked perceptual changes. Our approach permits a rapid, direct measurement of the consequences of image compression for human observers.",
author = "Christophe Charrier and Maloney, {Laurence T.} and Hocine Cherifi and Kenneth Knoblaunch",
year = "2007",
doi = "10.1364/JOSAA.24.003418",
language = "English (US)",
volume = "24",
pages = "3418--3426",
journal = "Journal of the Optical Society of America A: Optics and Image Science, and Vision",
issn = "0740-3232",
publisher = "The Optical Society",
number = "11",

}

TY - JOUR

T1 - Maximum likelihood difference scaling of image quality in compression-degraded images

AU - Charrier, Christophe

AU - Maloney, Laurence T.

AU - Cherifi, Hocine

AU - Knoblaunch, Kenneth

PY - 2007

Y1 - 2007

N2 - Lossy image compression techniques allow arbitrarily high compression rates but at the price of poor image quality. We applied maximum likelihood difference scaling to evaluate image quality of nine images, each compressed via vector quantization to ten different levels, within two different color spaces, RGB and CIE 1976 L*a*b*. In L*a*b* space, images could be compressed on average by 32% more than in RGB space, with little additional loss in quality. Further compression led to marked perceptual changes. Our approach permits a rapid, direct measurement of the consequences of image compression for human observers.

AB - Lossy image compression techniques allow arbitrarily high compression rates but at the price of poor image quality. We applied maximum likelihood difference scaling to evaluate image quality of nine images, each compressed via vector quantization to ten different levels, within two different color spaces, RGB and CIE 1976 L*a*b*. In L*a*b* space, images could be compressed on average by 32% more than in RGB space, with little additional loss in quality. Further compression led to marked perceptual changes. Our approach permits a rapid, direct measurement of the consequences of image compression for human observers.

UR - http://www.scopus.com/inward/record.url?scp=37549006779&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=37549006779&partnerID=8YFLogxK

U2 - 10.1364/JOSAA.24.003418

DO - 10.1364/JOSAA.24.003418

M3 - Article

VL - 24

SP - 3418

EP - 3426

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 - 0740-3232

IS - 11

ER -