On the choice of similarity measures for image retrieval by example

Jean Philippe Tarel, Sabri Boughorbel

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

5 Citations (Scopus)

Abstract

In image retrieval systems, a variety of simple similarity measures are used. The choice for one similarity measure or another is generally driven by an experimental comparison on a labeled database. The drawback of such an approach is that, while a large number of possible similarity measures can be tested, we do not know how to extend from the obtained results. However, the choice of a good similarity measure leads to noticeable better results. It is known that this choice is related to the variability of the images within the same class. Therefore, we propose a model of image retrieval systems and deduce a scheme for deriving the best similarity measure in a set of similarity measures, assuming a parametric model of the variability of feature vectors within the same class. An experimental validation of the model and the derived similarity measures is performed on synthetic ground-truth databases. Finally, from our experiments, we give several rules to follow for the design of ground-truth databases allowing reliable conclusions on the search of better similarity measures.

Original languageEnglish
Title of host publicationProceedings of the ACM International Multimedia Conference and Exhibition
Pages446-455
Number of pages10
Publication statusPublished - 2002
Externally publishedYes
Event10th International Conference of Multimedia - Juan les Pins, France
Duration: 1 Dec 20026 Dec 2002

Other

Other10th International Conference of Multimedia
CountryFrance
CityJuan les Pins
Period1/12/026/12/02

Fingerprint

Image retrieval
Experiments

ASJC Scopus subject areas

  • Computer Science(all)

Cite this

Tarel, J. P., & Boughorbel, S. (2002). On the choice of similarity measures for image retrieval by example. In Proceedings of the ACM International Multimedia Conference and Exhibition (pp. 446-455)

On the choice of similarity measures for image retrieval by example. / Tarel, Jean Philippe; Boughorbel, Sabri.

Proceedings of the ACM International Multimedia Conference and Exhibition. 2002. p. 446-455.

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

Tarel, JP & Boughorbel, S 2002, On the choice of similarity measures for image retrieval by example. in Proceedings of the ACM International Multimedia Conference and Exhibition. pp. 446-455, 10th International Conference of Multimedia, Juan les Pins, France, 1/12/02.
Tarel JP, Boughorbel S. On the choice of similarity measures for image retrieval by example. In Proceedings of the ACM International Multimedia Conference and Exhibition. 2002. p. 446-455
Tarel, Jean Philippe ; Boughorbel, Sabri. / On the choice of similarity measures for image retrieval by example. Proceedings of the ACM International Multimedia Conference and Exhibition. 2002. pp. 446-455
@inproceedings{45ae21c015fe480c9a97c877eb240938,
title = "On the choice of similarity measures for image retrieval by example",
abstract = "In image retrieval systems, a variety of simple similarity measures are used. The choice for one similarity measure or another is generally driven by an experimental comparison on a labeled database. The drawback of such an approach is that, while a large number of possible similarity measures can be tested, we do not know how to extend from the obtained results. However, the choice of a good similarity measure leads to noticeable better results. It is known that this choice is related to the variability of the images within the same class. Therefore, we propose a model of image retrieval systems and deduce a scheme for deriving the best similarity measure in a set of similarity measures, assuming a parametric model of the variability of feature vectors within the same class. An experimental validation of the model and the derived similarity measures is performed on synthetic ground-truth databases. Finally, from our experiments, we give several rules to follow for the design of ground-truth databases allowing reliable conclusions on the search of better similarity measures.",
author = "Tarel, {Jean Philippe} and Sabri Boughorbel",
year = "2002",
language = "English",
pages = "446--455",
booktitle = "Proceedings of the ACM International Multimedia Conference and Exhibition",

}

TY - GEN

T1 - On the choice of similarity measures for image retrieval by example

AU - Tarel, Jean Philippe

AU - Boughorbel, Sabri

PY - 2002

Y1 - 2002

N2 - In image retrieval systems, a variety of simple similarity measures are used. The choice for one similarity measure or another is generally driven by an experimental comparison on a labeled database. The drawback of such an approach is that, while a large number of possible similarity measures can be tested, we do not know how to extend from the obtained results. However, the choice of a good similarity measure leads to noticeable better results. It is known that this choice is related to the variability of the images within the same class. Therefore, we propose a model of image retrieval systems and deduce a scheme for deriving the best similarity measure in a set of similarity measures, assuming a parametric model of the variability of feature vectors within the same class. An experimental validation of the model and the derived similarity measures is performed on synthetic ground-truth databases. Finally, from our experiments, we give several rules to follow for the design of ground-truth databases allowing reliable conclusions on the search of better similarity measures.

AB - In image retrieval systems, a variety of simple similarity measures are used. The choice for one similarity measure or another is generally driven by an experimental comparison on a labeled database. The drawback of such an approach is that, while a large number of possible similarity measures can be tested, we do not know how to extend from the obtained results. However, the choice of a good similarity measure leads to noticeable better results. It is known that this choice is related to the variability of the images within the same class. Therefore, we propose a model of image retrieval systems and deduce a scheme for deriving the best similarity measure in a set of similarity measures, assuming a parametric model of the variability of feature vectors within the same class. An experimental validation of the model and the derived similarity measures is performed on synthetic ground-truth databases. Finally, from our experiments, we give several rules to follow for the design of ground-truth databases allowing reliable conclusions on the search of better similarity measures.

UR - http://www.scopus.com/inward/record.url?scp=0037700763&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0037700763&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0037700763

SP - 446

EP - 455

BT - Proceedings of the ACM International Multimedia Conference and Exhibition

ER -