Color-aware regularization for gradient domain image manipulation

Fanbo Deng, Seon Joo Kim, Yu Wing Tai, Michael S. Brown

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

2 Citations (Scopus)

Abstract

We propose a color-aware regularization for use with gradient domain image manipulation to avoid color shift artifacts. Our work is motivated by the observation that colors of objects in natural images typically follow distinct distributions in the color space. Conventional regularization methods ignore these distributions which can lead to undesirable colors appearing in the final output. Our approach uses an anisotropic Mahalanobis distance to control output colors to better fit original distributions. Our color-aware regularization is simple, easy to implement, and does not introduce significant computational overhead. To demonstrate the effectiveness of our method, we show the results with and without our color-aware regularization on three gradient domain tasks: gradient transfer, gradient boosting, and saliency sharpening.

Original languageEnglish
Title of host publicationComputer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers
Pages392-405
Number of pages14
EditionPART 4
DOIs
Publication statusPublished - 2013 Apr 11
Event11th Asian Conference on Computer Vision, ACCV 2012 - Daejeon, Korea, Republic of
Duration: 2012 Nov 52012 Nov 9

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 4
Volume7727 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other11th Asian Conference on Computer Vision, ACCV 2012
CountryKorea, Republic of
CityDaejeon
Period12/11/512/11/9

Fingerprint

Manipulation
Regularization
Gradient
Color
Mahalanobis Distance
Saliency
Output
Color Space
Regularization Method
Boosting
Distinct
Demonstrate

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Deng, F., Kim, S. J., Tai, Y. W., & Brown, M. S. (2013). Color-aware regularization for gradient domain image manipulation. In Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers (PART 4 ed., pp. 392-405). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7727 LNCS, No. PART 4). https://doi.org/10.1007/978-3-642-37447-0_30
Deng, Fanbo ; Kim, Seon Joo ; Tai, Yu Wing ; Brown, Michael S. / Color-aware regularization for gradient domain image manipulation. Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers. PART 4. ed. 2013. pp. 392-405 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 4).
@inproceedings{275fe5cf9aa84f049659838937791f3c,
title = "Color-aware regularization for gradient domain image manipulation",
abstract = "We propose a color-aware regularization for use with gradient domain image manipulation to avoid color shift artifacts. Our work is motivated by the observation that colors of objects in natural images typically follow distinct distributions in the color space. Conventional regularization methods ignore these distributions which can lead to undesirable colors appearing in the final output. Our approach uses an anisotropic Mahalanobis distance to control output colors to better fit original distributions. Our color-aware regularization is simple, easy to implement, and does not introduce significant computational overhead. To demonstrate the effectiveness of our method, we show the results with and without our color-aware regularization on three gradient domain tasks: gradient transfer, gradient boosting, and saliency sharpening.",
author = "Fanbo Deng and Kim, {Seon Joo} and Tai, {Yu Wing} and Brown, {Michael S.}",
year = "2013",
month = "4",
day = "11",
doi = "10.1007/978-3-642-37447-0_30",
language = "English",
isbn = "9783642374463",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 4",
pages = "392--405",
booktitle = "Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers",
edition = "PART 4",

}

Deng, F, Kim, SJ, Tai, YW & Brown, MS 2013, Color-aware regularization for gradient domain image manipulation. in Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers. PART 4 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 4, vol. 7727 LNCS, pp. 392-405, 11th Asian Conference on Computer Vision, ACCV 2012, Daejeon, Korea, Republic of, 12/11/5. https://doi.org/10.1007/978-3-642-37447-0_30

Color-aware regularization for gradient domain image manipulation. / Deng, Fanbo; Kim, Seon Joo; Tai, Yu Wing; Brown, Michael S.

Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers. PART 4. ed. 2013. p. 392-405 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7727 LNCS, No. PART 4).

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

TY - GEN

T1 - Color-aware regularization for gradient domain image manipulation

AU - Deng, Fanbo

AU - Kim, Seon Joo

AU - Tai, Yu Wing

AU - Brown, Michael S.

PY - 2013/4/11

Y1 - 2013/4/11

N2 - We propose a color-aware regularization for use with gradient domain image manipulation to avoid color shift artifacts. Our work is motivated by the observation that colors of objects in natural images typically follow distinct distributions in the color space. Conventional regularization methods ignore these distributions which can lead to undesirable colors appearing in the final output. Our approach uses an anisotropic Mahalanobis distance to control output colors to better fit original distributions. Our color-aware regularization is simple, easy to implement, and does not introduce significant computational overhead. To demonstrate the effectiveness of our method, we show the results with and without our color-aware regularization on three gradient domain tasks: gradient transfer, gradient boosting, and saliency sharpening.

AB - We propose a color-aware regularization for use with gradient domain image manipulation to avoid color shift artifacts. Our work is motivated by the observation that colors of objects in natural images typically follow distinct distributions in the color space. Conventional regularization methods ignore these distributions which can lead to undesirable colors appearing in the final output. Our approach uses an anisotropic Mahalanobis distance to control output colors to better fit original distributions. Our color-aware regularization is simple, easy to implement, and does not introduce significant computational overhead. To demonstrate the effectiveness of our method, we show the results with and without our color-aware regularization on three gradient domain tasks: gradient transfer, gradient boosting, and saliency sharpening.

UR - http://www.scopus.com/inward/record.url?scp=84875907068&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84875907068&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-37447-0_30

DO - 10.1007/978-3-642-37447-0_30

M3 - Conference contribution

AN - SCOPUS:84875907068

SN - 9783642374463

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 392

EP - 405

BT - Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers

ER -

Deng F, Kim SJ, Tai YW, Brown MS. Color-aware regularization for gradient domain image manipulation. In Computer Vision, ACCV 2012 - 11th Asian Conference on Computer Vision, Revised Selected Papers. PART 4 ed. 2013. p. 392-405. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 4). https://doi.org/10.1007/978-3-642-37447-0_30