Color and Color Constancy in a Translation Model for Object Recognition

Jacobus J Barnard, Prasad Gabbur

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

10 Citations (Scopus)

Abstract

Color is of interest to those working in computer vision largely because it is assumed to be helpful for recognition. This assumption has driven much work in color based image indexing, and computational color constancy. However, in many ways, indexing is a poor model for recognition. In this paper we use a recently developed statistical model of recognition which learns to link image region features with words, based on a large unstructured data set. The system is general in that it learns what is recognizable given the data. It also supports a principled testing paradigm which we exploit here to evaluate the use of color. In particular, we look at color space choice, degradation due to illumination change, and dealing with this degradation. We evaluate two general approaches to dealing with this color constancy problem. Specifically we address whether it is better to build color variation due to illumination into a recognition system, or, instead, apply color constancy preprocessing to images before they are processed by the recognition system.

Original languageEnglish (US)
Title of host publicationFinal Program and Proceedings - IS and T/SID Color Imaging Conference
Pages364-369
Number of pages6
StatePublished - 2003
EventFinal Program and Proceedings of the IS and T and SID - 11th Color Imaging Conference: Color Science and Engineering:Systems, Technologies, Applications - Scottdale, AZ, United States
Duration: Nov 3 2003Nov 7 2003

Other

OtherFinal Program and Proceedings of the IS and T and SID - 11th Color Imaging Conference: Color Science and Engineering:Systems, Technologies, Applications
CountryUnited States
CityScottdale, AZ
Period11/3/0311/7/03

Fingerprint

Object recognition
Color
Lighting
Degradation
Computer vision
Testing

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Barnard, J. J., & Gabbur, P. (2003). Color and Color Constancy in a Translation Model for Object Recognition. In Final Program and Proceedings - IS and T/SID Color Imaging Conference (pp. 364-369)

Color and Color Constancy in a Translation Model for Object Recognition. / Barnard, Jacobus J; Gabbur, Prasad.

Final Program and Proceedings - IS and T/SID Color Imaging Conference. 2003. p. 364-369.

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

Barnard, JJ & Gabbur, P 2003, Color and Color Constancy in a Translation Model for Object Recognition. in Final Program and Proceedings - IS and T/SID Color Imaging Conference. pp. 364-369, Final Program and Proceedings of the IS and T and SID - 11th Color Imaging Conference: Color Science and Engineering:Systems, Technologies, Applications, Scottdale, AZ, United States, 11/3/03.
Barnard JJ, Gabbur P. Color and Color Constancy in a Translation Model for Object Recognition. In Final Program and Proceedings - IS and T/SID Color Imaging Conference. 2003. p. 364-369
Barnard, Jacobus J ; Gabbur, Prasad. / Color and Color Constancy in a Translation Model for Object Recognition. Final Program and Proceedings - IS and T/SID Color Imaging Conference. 2003. pp. 364-369
@inproceedings{f19e9672c0924fa68c175a7ea587f081,
title = "Color and Color Constancy in a Translation Model for Object Recognition",
abstract = "Color is of interest to those working in computer vision largely because it is assumed to be helpful for recognition. This assumption has driven much work in color based image indexing, and computational color constancy. However, in many ways, indexing is a poor model for recognition. In this paper we use a recently developed statistical model of recognition which learns to link image region features with words, based on a large unstructured data set. The system is general in that it learns what is recognizable given the data. It also supports a principled testing paradigm which we exploit here to evaluate the use of color. In particular, we look at color space choice, degradation due to illumination change, and dealing with this degradation. We evaluate two general approaches to dealing with this color constancy problem. Specifically we address whether it is better to build color variation due to illumination into a recognition system, or, instead, apply color constancy preprocessing to images before they are processed by the recognition system.",
author = "Barnard, {Jacobus J} and Prasad Gabbur",
year = "2003",
language = "English (US)",
pages = "364--369",
booktitle = "Final Program and Proceedings - IS and T/SID Color Imaging Conference",

}

TY - GEN

T1 - Color and Color Constancy in a Translation Model for Object Recognition

AU - Barnard, Jacobus J

AU - Gabbur, Prasad

PY - 2003

Y1 - 2003

N2 - Color is of interest to those working in computer vision largely because it is assumed to be helpful for recognition. This assumption has driven much work in color based image indexing, and computational color constancy. However, in many ways, indexing is a poor model for recognition. In this paper we use a recently developed statistical model of recognition which learns to link image region features with words, based on a large unstructured data set. The system is general in that it learns what is recognizable given the data. It also supports a principled testing paradigm which we exploit here to evaluate the use of color. In particular, we look at color space choice, degradation due to illumination change, and dealing with this degradation. We evaluate two general approaches to dealing with this color constancy problem. Specifically we address whether it is better to build color variation due to illumination into a recognition system, or, instead, apply color constancy preprocessing to images before they are processed by the recognition system.

AB - Color is of interest to those working in computer vision largely because it is assumed to be helpful for recognition. This assumption has driven much work in color based image indexing, and computational color constancy. However, in many ways, indexing is a poor model for recognition. In this paper we use a recently developed statistical model of recognition which learns to link image region features with words, based on a large unstructured data set. The system is general in that it learns what is recognizable given the data. It also supports a principled testing paradigm which we exploit here to evaluate the use of color. In particular, we look at color space choice, degradation due to illumination change, and dealing with this degradation. We evaluate two general approaches to dealing with this color constancy problem. Specifically we address whether it is better to build color variation due to illumination into a recognition system, or, instead, apply color constancy preprocessing to images before they are processed by the recognition system.

UR - http://www.scopus.com/inward/record.url?scp=16244404803&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=16244404803&partnerID=8YFLogxK

M3 - Conference contribution

SP - 364

EP - 369

BT - Final Program and Proceedings - IS and T/SID Color Imaging Conference

ER -