Vector space model adaptation and pseudo relevance feedback for content-based image retrieval

H. Karamti, M. Tmar, M. Visani, T. Urruty, F. Gargouri

Research output: Contribution to journalArticleResearchpeer-review

2 Citations (Scopus)

Abstract

Image retrieval is an important problem for researchers in computer vision and content-based image retrieval (CBIR) fields. Over the last decades, many image retrieval systems were based on image representation as a set of extracted low-level features such as color, texture and shape. Then, systems calculate similarity metrics between features in order to find similar images to a query image. The disadvantage of this approach is that images visually and semantically different may be similar in the low level feature space. So, it is necessary to develop tools to optimize retrieval of information. Integration of vector space models is one solution to improve the performance of image retrieval. In this paper, we present an efficient and effective retrieval framework which includes a vectorization technique combined with a pseudo relevance model. The idea is to transform any similarity matching model (between images) to a vector space model providing a score. A study on several methodologies to obtain the vectorization is presented. Some experiments have been undertaken on Wang, Oxford5k and Inria Holidays datasets to show the performance of our proposed framework.

Original languageEnglish
Pages (from-to)5475-5501
Number of pages27
JournalMultimedia Tools and Applications
Volume77
Issue number5
DOIs
Publication statusPublished - 1 Mar 2018

Fingerprint

Image retrieval
Vector spaces
Feedback
Computer vision
Textures
Color
Experiments

Keywords

  • CBIR
  • Late fusion
  • Neural network
  • Pseudo-relevance feedback
  • Vector space model
  • Vectorization

Cite this

Karamti, H. ; Tmar, M. ; Visani, M. ; Urruty, T. ; Gargouri, F. / Vector space model adaptation and pseudo relevance feedback for content-based image retrieval. In: Multimedia Tools and Applications. 2018 ; Vol. 77, No. 5. pp. 5475-5501.
@article{97635c6d03724bea99e1ec9ff3045eff,
title = "Vector space model adaptation and pseudo relevance feedback for content-based image retrieval",
abstract = "Image retrieval is an important problem for researchers in computer vision and content-based image retrieval (CBIR) fields. Over the last decades, many image retrieval systems were based on image representation as a set of extracted low-level features such as color, texture and shape. Then, systems calculate similarity metrics between features in order to find similar images to a query image. The disadvantage of this approach is that images visually and semantically different may be similar in the low level feature space. So, it is necessary to develop tools to optimize retrieval of information. Integration of vector space models is one solution to improve the performance of image retrieval. In this paper, we present an efficient and effective retrieval framework which includes a vectorization technique combined with a pseudo relevance model. The idea is to transform any similarity matching model (between images) to a vector space model providing a score. A study on several methodologies to obtain the vectorization is presented. Some experiments have been undertaken on Wang, Oxford5k and Inria Holidays datasets to show the performance of our proposed framework.",
keywords = "CBIR, Late fusion, Neural network, Pseudo-relevance feedback, Vector space model, Vectorization",
author = "H. Karamti and M. Tmar and M. Visani and T. Urruty and F. Gargouri",
year = "2018",
month = "3",
day = "1",
doi = "10.1007/s11042-017-4463-x",
language = "English",
volume = "77",
pages = "5475--5501",
journal = "Multimedia Tools and Applications",
issn = "1380-7501",
publisher = "Springer Netherlands",
number = "5",

}

Vector space model adaptation and pseudo relevance feedback for content-based image retrieval. / Karamti, H.; Tmar, M.; Visani, M.; Urruty, T.; Gargouri, F.

In: Multimedia Tools and Applications, Vol. 77, No. 5, 01.03.2018, p. 5475-5501.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Vector space model adaptation and pseudo relevance feedback for content-based image retrieval

AU - Karamti, H.

AU - Tmar, M.

AU - Visani, M.

AU - Urruty, T.

AU - Gargouri, F.

PY - 2018/3/1

Y1 - 2018/3/1

N2 - Image retrieval is an important problem for researchers in computer vision and content-based image retrieval (CBIR) fields. Over the last decades, many image retrieval systems were based on image representation as a set of extracted low-level features such as color, texture and shape. Then, systems calculate similarity metrics between features in order to find similar images to a query image. The disadvantage of this approach is that images visually and semantically different may be similar in the low level feature space. So, it is necessary to develop tools to optimize retrieval of information. Integration of vector space models is one solution to improve the performance of image retrieval. In this paper, we present an efficient and effective retrieval framework which includes a vectorization technique combined with a pseudo relevance model. The idea is to transform any similarity matching model (between images) to a vector space model providing a score. A study on several methodologies to obtain the vectorization is presented. Some experiments have been undertaken on Wang, Oxford5k and Inria Holidays datasets to show the performance of our proposed framework.

AB - Image retrieval is an important problem for researchers in computer vision and content-based image retrieval (CBIR) fields. Over the last decades, many image retrieval systems were based on image representation as a set of extracted low-level features such as color, texture and shape. Then, systems calculate similarity metrics between features in order to find similar images to a query image. The disadvantage of this approach is that images visually and semantically different may be similar in the low level feature space. So, it is necessary to develop tools to optimize retrieval of information. Integration of vector space models is one solution to improve the performance of image retrieval. In this paper, we present an efficient and effective retrieval framework which includes a vectorization technique combined with a pseudo relevance model. The idea is to transform any similarity matching model (between images) to a vector space model providing a score. A study on several methodologies to obtain the vectorization is presented. Some experiments have been undertaken on Wang, Oxford5k and Inria Holidays datasets to show the performance of our proposed framework.

KW - CBIR

KW - Late fusion

KW - Neural network

KW - Pseudo-relevance feedback

KW - Vector space model

KW - Vectorization

UR - http://www.scopus.com/inward/record.url?scp=85015242361&partnerID=8YFLogxK

U2 - 10.1007/s11042-017-4463-x

DO - 10.1007/s11042-017-4463-x

M3 - Article

VL - 77

SP - 5475

EP - 5501

JO - Multimedia Tools and Applications

JF - Multimedia Tools and Applications

SN - 1380-7501

IS - 5

ER -