Illumination normalisation using convolutional neural network with application to face recognition

Y. H. Kim, H. Kim, S. W. Kim, H. Y. Kim, Sung-Jea Ko

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

A novel illumination normalisation (IN) method using a convolutional neural network (CNN) is proposed. The proposed network is composed of the local pattern extraction (LPE) and illumination elimination (IE) layers. The LPE layers model the relationships between the pixels in each local region in order to handle various types of local shadow and shading in the face image. Based on the commonly used assumption about the illumination field, the IE layers generate illuminationinsensitive ratio images by calculating the ratio between the output pairs produced from the LPE layers. The final feature map obtained by combining the ratio images can possess an improved discriminative ability for face recognition (FR). For training the proposed network, the results produced by the Weber fraction-based IN methods as ground truths are utilised. The experimental results demonstrate that the proposed network performs better in terms of FR accuracy compared with the conventional non-CNN-based method and it can be combined with any CNN-based face classifier.

Original languageEnglish
Pages (from-to)399-401
Number of pages3
JournalElectronics Letters
Volume53
Issue number6
DOIs
Publication statusPublished - 2017 Mar 16

Fingerprint

Face recognition
Lighting
Neural networks
Classifiers
Pixels

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Cite this

Illumination normalisation using convolutional neural network with application to face recognition. / Kim, Y. H.; Kim, H.; Kim, S. W.; Kim, H. Y.; Ko, Sung-Jea.

In: Electronics Letters, Vol. 53, No. 6, 16.03.2017, p. 399-401.

Research output: Contribution to journalArticle

Kim, Y. H. ; Kim, H. ; Kim, S. W. ; Kim, H. Y. ; Ko, Sung-Jea. / Illumination normalisation using convolutional neural network with application to face recognition. In: Electronics Letters. 2017 ; Vol. 53, No. 6. pp. 399-401.
@article{f9fca5aca8bd44a9bcb916fd80ecbb1b,
title = "Illumination normalisation using convolutional neural network with application to face recognition",
abstract = "A novel illumination normalisation (IN) method using a convolutional neural network (CNN) is proposed. The proposed network is composed of the local pattern extraction (LPE) and illumination elimination (IE) layers. The LPE layers model the relationships between the pixels in each local region in order to handle various types of local shadow and shading in the face image. Based on the commonly used assumption about the illumination field, the IE layers generate illuminationinsensitive ratio images by calculating the ratio between the output pairs produced from the LPE layers. The final feature map obtained by combining the ratio images can possess an improved discriminative ability for face recognition (FR). For training the proposed network, the results produced by the Weber fraction-based IN methods as ground truths are utilised. The experimental results demonstrate that the proposed network performs better in terms of FR accuracy compared with the conventional non-CNN-based method and it can be combined with any CNN-based face classifier.",
author = "Kim, {Y. H.} and H. Kim and Kim, {S. W.} and Kim, {H. Y.} and Sung-Jea Ko",
year = "2017",
month = "3",
day = "16",
doi = "10.1049/el.2017.0023",
language = "English",
volume = "53",
pages = "399--401",
journal = "Electronics Letters",
issn = "0013-5194",
publisher = "Institution of Engineering and Technology",
number = "6",

}

TY - JOUR

T1 - Illumination normalisation using convolutional neural network with application to face recognition

AU - Kim, Y. H.

AU - Kim, H.

AU - Kim, S. W.

AU - Kim, H. Y.

AU - Ko, Sung-Jea

PY - 2017/3/16

Y1 - 2017/3/16

N2 - A novel illumination normalisation (IN) method using a convolutional neural network (CNN) is proposed. The proposed network is composed of the local pattern extraction (LPE) and illumination elimination (IE) layers. The LPE layers model the relationships between the pixels in each local region in order to handle various types of local shadow and shading in the face image. Based on the commonly used assumption about the illumination field, the IE layers generate illuminationinsensitive ratio images by calculating the ratio between the output pairs produced from the LPE layers. The final feature map obtained by combining the ratio images can possess an improved discriminative ability for face recognition (FR). For training the proposed network, the results produced by the Weber fraction-based IN methods as ground truths are utilised. The experimental results demonstrate that the proposed network performs better in terms of FR accuracy compared with the conventional non-CNN-based method and it can be combined with any CNN-based face classifier.

AB - A novel illumination normalisation (IN) method using a convolutional neural network (CNN) is proposed. The proposed network is composed of the local pattern extraction (LPE) and illumination elimination (IE) layers. The LPE layers model the relationships between the pixels in each local region in order to handle various types of local shadow and shading in the face image. Based on the commonly used assumption about the illumination field, the IE layers generate illuminationinsensitive ratio images by calculating the ratio between the output pairs produced from the LPE layers. The final feature map obtained by combining the ratio images can possess an improved discriminative ability for face recognition (FR). For training the proposed network, the results produced by the Weber fraction-based IN methods as ground truths are utilised. The experimental results demonstrate that the proposed network performs better in terms of FR accuracy compared with the conventional non-CNN-based method and it can be combined with any CNN-based face classifier.

UR - http://www.scopus.com/inward/record.url?scp=85015761161&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85015761161&partnerID=8YFLogxK

U2 - 10.1049/el.2017.0023

DO - 10.1049/el.2017.0023

M3 - Article

AN - SCOPUS:85015761161

VL - 53

SP - 399

EP - 401

JO - Electronics Letters

JF - Electronics Letters

SN - 0013-5194

IS - 6

ER -