A new vector space model for image retrieval

Hanen Karamti, Mohamed Tmar, Faiez Gargouri

Research output: Contribution to journalConference articleResearchpeer-review

Abstract

The rapid development of digitization and data storage techniques resulted in images volume increase. In order to cope with this increasing amount of informations, it is necessary to develop tools to accelerate and facilitate the access to information and to ensure the relevance of information available to users. These tools must minimize the problems related to the image indexing used to represent content query information. In this paper, we present a new retrieval model called vectorization. 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 Oxford5k and Inria Holidays datasets to show the performance of our proposed system.

Original languageEnglish
Pages (from-to)771-779
Number of pages9
JournalProcedia Computer Science
Volume112
DOIs
Publication statusPublished - 1 Jan 2017
Event21st International Conference on Knowledge - Based and Intelligent Information and Engineering Systems, KES 2017 - Marseille, France
Duration: 6 Sep 20178 Sep 2017

Fingerprint

Image retrieval
Vector spaces
Analog to digital conversion
Data storage equipment
Experiments

Keywords

  • CBIR
  • deep learning
  • late fusion
  • neural network
  • vector space model
  • Vectorization

Cite this

Karamti, Hanen ; Tmar, Mohamed ; Gargouri, Faiez. / A new vector space model for image retrieval. In: Procedia Computer Science. 2017 ; Vol. 112. pp. 771-779.
@article{b88f76f44245441aaa90aa63ffa6f754,
title = "A new vector space model for image retrieval",
abstract = "The rapid development of digitization and data storage techniques resulted in images volume increase. In order to cope with this increasing amount of informations, it is necessary to develop tools to accelerate and facilitate the access to information and to ensure the relevance of information available to users. These tools must minimize the problems related to the image indexing used to represent content query information. In this paper, we present a new retrieval model called vectorization. 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 Oxford5k and Inria Holidays datasets to show the performance of our proposed system.",
keywords = "CBIR, deep learning, late fusion, neural network, vector space model, Vectorization",
author = "Hanen Karamti and Mohamed Tmar and Faiez Gargouri",
year = "2017",
month = "1",
day = "1",
doi = "10.1016/j.procs.2017.08.202",
language = "English",
volume = "112",
pages = "771--779",
journal = "Procedia Computer Science",
issn = "1877-0509",
publisher = "Elsevier BV",

}

A new vector space model for image retrieval. / Karamti, Hanen; Tmar, Mohamed; Gargouri, Faiez.

In: Procedia Computer Science, Vol. 112, 01.01.2017, p. 771-779.

Research output: Contribution to journalConference articleResearchpeer-review

TY - JOUR

T1 - A new vector space model for image retrieval

AU - Karamti, Hanen

AU - Tmar, Mohamed

AU - Gargouri, Faiez

PY - 2017/1/1

Y1 - 2017/1/1

N2 - The rapid development of digitization and data storage techniques resulted in images volume increase. In order to cope with this increasing amount of informations, it is necessary to develop tools to accelerate and facilitate the access to information and to ensure the relevance of information available to users. These tools must minimize the problems related to the image indexing used to represent content query information. In this paper, we present a new retrieval model called vectorization. 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 Oxford5k and Inria Holidays datasets to show the performance of our proposed system.

AB - The rapid development of digitization and data storage techniques resulted in images volume increase. In order to cope with this increasing amount of informations, it is necessary to develop tools to accelerate and facilitate the access to information and to ensure the relevance of information available to users. These tools must minimize the problems related to the image indexing used to represent content query information. In this paper, we present a new retrieval model called vectorization. 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 Oxford5k and Inria Holidays datasets to show the performance of our proposed system.

KW - CBIR

KW - deep learning

KW - late fusion

KW - neural network

KW - vector space model

KW - Vectorization

UR - http://www.scopus.com/inward/record.url?scp=85032333522&partnerID=8YFLogxK

U2 - 10.1016/j.procs.2017.08.202

DO - 10.1016/j.procs.2017.08.202

M3 - Conference article

VL - 112

SP - 771

EP - 779

JO - Procedia Computer Science

JF - Procedia Computer Science

SN - 1877-0509

ER -