Manifold-ranking based image retrieval

Jingrui He, Mingjing Li, Hong Jiang Zhang, Hanghang Tong, Changshui Zhang

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

281 Citations (Scopus)

Abstract

In this paper, we propose a novel transductive learning framework named manifold-ranking based image retrieval (MRBIR). Given a query image, MRBIR first makes use of a manifold ranking algorithm to explore the relationship among all the data points in the feature space, and then measures relevance between the query and all the images in the database accordingly, which is different from traditional similarity metrics based on pair-wise distance. In relevance feedback, if only positive examples are available, they are added to the query set to improve the retrieval result; if examples of both labels can be obtained, MRBIR discriminately spreads the ranking scores of positive and negative examples, considering the asymmetry between these two types of images. Furthermore, three active learning methods are incorporated into MRBIR, which select images in each round of relevance feedback according to different principles, aiming to maximally improve the ranking result. Experimental results on a general-purpose image database show that MRBIR attains a significant improvement over existing systems from all aspects.

Original languageEnglish (US)
Title of host publicationACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia
Pages9-16
Number of pages8
StatePublished - 2004
Externally publishedYes
EventACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia - New York, NY, United States
Duration: Oct 10 2004Oct 16 2004

Other

OtherACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia
CountryUnited States
CityNew York, NY
Period10/10/0410/16/04

Fingerprint

Image retrieval
Feedback
Labels

Keywords

  • Active learning
  • Image retrieval
  • Manifold ranking
  • Relevance feedback

ASJC Scopus subject areas

  • Engineering(all)

Cite this

He, J., Li, M., Zhang, H. J., Tong, H., & Zhang, C. (2004). Manifold-ranking based image retrieval. In ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia (pp. 9-16)

Manifold-ranking based image retrieval. / He, Jingrui; Li, Mingjing; Zhang, Hong Jiang; Tong, Hanghang; Zhang, Changshui.

ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia. 2004. p. 9-16.

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

He, J, Li, M, Zhang, HJ, Tong, H & Zhang, C 2004, Manifold-ranking based image retrieval. in ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia. pp. 9-16, ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia, New York, NY, United States, 10/10/04.
He J, Li M, Zhang HJ, Tong H, Zhang C. Manifold-ranking based image retrieval. In ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia. 2004. p. 9-16
He, Jingrui ; Li, Mingjing ; Zhang, Hong Jiang ; Tong, Hanghang ; Zhang, Changshui. / Manifold-ranking based image retrieval. ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia. 2004. pp. 9-16
@inproceedings{c83278e143ce4b09b478a0ef4500cf69,
title = "Manifold-ranking based image retrieval",
abstract = "In this paper, we propose a novel transductive learning framework named manifold-ranking based image retrieval (MRBIR). Given a query image, MRBIR first makes use of a manifold ranking algorithm to explore the relationship among all the data points in the feature space, and then measures relevance between the query and all the images in the database accordingly, which is different from traditional similarity metrics based on pair-wise distance. In relevance feedback, if only positive examples are available, they are added to the query set to improve the retrieval result; if examples of both labels can be obtained, MRBIR discriminately spreads the ranking scores of positive and negative examples, considering the asymmetry between these two types of images. Furthermore, three active learning methods are incorporated into MRBIR, which select images in each round of relevance feedback according to different principles, aiming to maximally improve the ranking result. Experimental results on a general-purpose image database show that MRBIR attains a significant improvement over existing systems from all aspects.",
keywords = "Active learning, Image retrieval, Manifold ranking, Relevance feedback",
author = "Jingrui He and Mingjing Li and Zhang, {Hong Jiang} and Hanghang Tong and Changshui Zhang",
year = "2004",
language = "English (US)",
isbn = "1581138938",
pages = "9--16",
booktitle = "ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia",

}

TY - GEN

T1 - Manifold-ranking based image retrieval

AU - He, Jingrui

AU - Li, Mingjing

AU - Zhang, Hong Jiang

AU - Tong, Hanghang

AU - Zhang, Changshui

PY - 2004

Y1 - 2004

N2 - In this paper, we propose a novel transductive learning framework named manifold-ranking based image retrieval (MRBIR). Given a query image, MRBIR first makes use of a manifold ranking algorithm to explore the relationship among all the data points in the feature space, and then measures relevance between the query and all the images in the database accordingly, which is different from traditional similarity metrics based on pair-wise distance. In relevance feedback, if only positive examples are available, they are added to the query set to improve the retrieval result; if examples of both labels can be obtained, MRBIR discriminately spreads the ranking scores of positive and negative examples, considering the asymmetry between these two types of images. Furthermore, three active learning methods are incorporated into MRBIR, which select images in each round of relevance feedback according to different principles, aiming to maximally improve the ranking result. Experimental results on a general-purpose image database show that MRBIR attains a significant improvement over existing systems from all aspects.

AB - In this paper, we propose a novel transductive learning framework named manifold-ranking based image retrieval (MRBIR). Given a query image, MRBIR first makes use of a manifold ranking algorithm to explore the relationship among all the data points in the feature space, and then measures relevance between the query and all the images in the database accordingly, which is different from traditional similarity metrics based on pair-wise distance. In relevance feedback, if only positive examples are available, they are added to the query set to improve the retrieval result; if examples of both labels can be obtained, MRBIR discriminately spreads the ranking scores of positive and negative examples, considering the asymmetry between these two types of images. Furthermore, three active learning methods are incorporated into MRBIR, which select images in each round of relevance feedback according to different principles, aiming to maximally improve the ranking result. Experimental results on a general-purpose image database show that MRBIR attains a significant improvement over existing systems from all aspects.

KW - Active learning

KW - Image retrieval

KW - Manifold ranking

KW - Relevance feedback

UR - http://www.scopus.com/inward/record.url?scp=13444263430&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=13444263430&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:13444263430

SN - 1581138938

SN - 9781581138931

SP - 9

EP - 16

BT - ACM Multimedia 2004 - proceedings of the 12th ACM International Conference on Multimedia

ER -