Generalized histogram intersection kernel for image recognition

Sabri Boughorbel, Jean Philippe Tarel, Nozha Boujemaa

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

65 Citations (Scopus)

Abstract

Histogram Intersection (HI) kernel has been recently introduced for image recognition tasks. The HI kernel is proved to be positive definite and thus can be used in Support Vector Machine (SVM) based recognition. Experimentally, it also leads to good recognition performances. However, its derivation applies only for binary strings such as color histograms computed on equally sized images. In this paper, we propose a new kernel, which we named Generalized Histogram Intersection (GHI) kernel, since it applies in a much larger variety of contexts. First, an original derivation of the positive definiteness of the GHI kernel is proposed in the general case. As a consequence, vectors of real values can be used, and the images no longer need to have the same size. Second, a hyper-parameter is added, compared to the HI kernel, which allows us to better tune the kernel model to particular databases. We present experiments which prove that the GHI kernel outperforms the simple HI kernel in a simple recognition task. Comparisons with other well-known kernels are also provided.

Original languageEnglish
Title of host publicationIEEE International Conference on Image Processing 2005, ICIP 2005
Pages161-164
Number of pages4
Volume3
DOIs
Publication statusPublished - 2005
Externally publishedYes
EventIEEE International Conference on Image Processing 2005, ICIP 2005 - Genova, Italy
Duration: 11 Sep 200514 Sep 2005

Other

OtherIEEE International Conference on Image Processing 2005, ICIP 2005
CountryItaly
CityGenova
Period11/9/0514/9/05

Fingerprint

Image recognition
Support vector machines
Color

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Boughorbel, S., Tarel, J. P., & Boujemaa, N. (2005). Generalized histogram intersection kernel for image recognition. In IEEE International Conference on Image Processing 2005, ICIP 2005 (Vol. 3, pp. 161-164). [1530353] https://doi.org/10.1109/ICIP.2005.1530353

Generalized histogram intersection kernel for image recognition. / Boughorbel, Sabri; Tarel, Jean Philippe; Boujemaa, Nozha.

IEEE International Conference on Image Processing 2005, ICIP 2005. Vol. 3 2005. p. 161-164 1530353.

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

Boughorbel, S, Tarel, JP & Boujemaa, N 2005, Generalized histogram intersection kernel for image recognition. in IEEE International Conference on Image Processing 2005, ICIP 2005. vol. 3, 1530353, pp. 161-164, IEEE International Conference on Image Processing 2005, ICIP 2005, Genova, Italy, 11/9/05. https://doi.org/10.1109/ICIP.2005.1530353
Boughorbel S, Tarel JP, Boujemaa N. Generalized histogram intersection kernel for image recognition. In IEEE International Conference on Image Processing 2005, ICIP 2005. Vol. 3. 2005. p. 161-164. 1530353 https://doi.org/10.1109/ICIP.2005.1530353
Boughorbel, Sabri ; Tarel, Jean Philippe ; Boujemaa, Nozha. / Generalized histogram intersection kernel for image recognition. IEEE International Conference on Image Processing 2005, ICIP 2005. Vol. 3 2005. pp. 161-164
@inproceedings{1938646c6f64427a9c1a9a880a03bcf1,
title = "Generalized histogram intersection kernel for image recognition",
abstract = "Histogram Intersection (HI) kernel has been recently introduced for image recognition tasks. The HI kernel is proved to be positive definite and thus can be used in Support Vector Machine (SVM) based recognition. Experimentally, it also leads to good recognition performances. However, its derivation applies only for binary strings such as color histograms computed on equally sized images. In this paper, we propose a new kernel, which we named Generalized Histogram Intersection (GHI) kernel, since it applies in a much larger variety of contexts. First, an original derivation of the positive definiteness of the GHI kernel is proposed in the general case. As a consequence, vectors of real values can be used, and the images no longer need to have the same size. Second, a hyper-parameter is added, compared to the HI kernel, which allows us to better tune the kernel model to particular databases. We present experiments which prove that the GHI kernel outperforms the simple HI kernel in a simple recognition task. Comparisons with other well-known kernels are also provided.",
author = "Sabri Boughorbel and Tarel, {Jean Philippe} and Nozha Boujemaa",
year = "2005",
doi = "10.1109/ICIP.2005.1530353",
language = "English",
isbn = "0780391349",
volume = "3",
pages = "161--164",
booktitle = "IEEE International Conference on Image Processing 2005, ICIP 2005",

}

TY - GEN

T1 - Generalized histogram intersection kernel for image recognition

AU - Boughorbel, Sabri

AU - Tarel, Jean Philippe

AU - Boujemaa, Nozha

PY - 2005

Y1 - 2005

N2 - Histogram Intersection (HI) kernel has been recently introduced for image recognition tasks. The HI kernel is proved to be positive definite and thus can be used in Support Vector Machine (SVM) based recognition. Experimentally, it also leads to good recognition performances. However, its derivation applies only for binary strings such as color histograms computed on equally sized images. In this paper, we propose a new kernel, which we named Generalized Histogram Intersection (GHI) kernel, since it applies in a much larger variety of contexts. First, an original derivation of the positive definiteness of the GHI kernel is proposed in the general case. As a consequence, vectors of real values can be used, and the images no longer need to have the same size. Second, a hyper-parameter is added, compared to the HI kernel, which allows us to better tune the kernel model to particular databases. We present experiments which prove that the GHI kernel outperforms the simple HI kernel in a simple recognition task. Comparisons with other well-known kernels are also provided.

AB - Histogram Intersection (HI) kernel has been recently introduced for image recognition tasks. The HI kernel is proved to be positive definite and thus can be used in Support Vector Machine (SVM) based recognition. Experimentally, it also leads to good recognition performances. However, its derivation applies only for binary strings such as color histograms computed on equally sized images. In this paper, we propose a new kernel, which we named Generalized Histogram Intersection (GHI) kernel, since it applies in a much larger variety of contexts. First, an original derivation of the positive definiteness of the GHI kernel is proposed in the general case. As a consequence, vectors of real values can be used, and the images no longer need to have the same size. Second, a hyper-parameter is added, compared to the HI kernel, which allows us to better tune the kernel model to particular databases. We present experiments which prove that the GHI kernel outperforms the simple HI kernel in a simple recognition task. Comparisons with other well-known kernels are also provided.

UR - http://www.scopus.com/inward/record.url?scp=33749268008&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33749268008&partnerID=8YFLogxK

U2 - 10.1109/ICIP.2005.1530353

DO - 10.1109/ICIP.2005.1530353

M3 - Conference contribution

SN - 0780391349

SN - 9780780391345

VL - 3

SP - 161

EP - 164

BT - IEEE International Conference on Image Processing 2005, ICIP 2005

ER -