Retinex-based illumination normalization using class-based illumination subspace for robust face recognition

Seung Wook Kim, June Young Jung, Cheol Hwan Yoo, Sung-Jea Ko

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Recent illumination normalization (IN) methods first decompose a face image into a reflectance (R)-image having a lighting-invariant characteristic and an illuminance (I)-image including shading and shadowing effects. An illumination-normalized I-image is then obtained by eliminating the lighting-dependent image variations (LDIV) from the I-image. Finally, the normalized I-and R-images are recombined for face recognition (FR). However, the decomposed-reflectance is often contaminated with the lighting effects. Moreover, the lighting normalization tends to remove the valuable discriminant information in the I-image. To address these problems, we employ the local edge-preserving filter to generate the R-image whereby the lighting-invariant information is well preserved. In addition, we propose a subspace-based IN method that can retain the large facial-structure in the I-image. To construct the proposed subspace, we calculate the LDIV within the same class of people from the training database of face images. Then, we apply the singular value decomposition to the calculated LDIV to obtain the basis images of the subspace. By projecting the I-image onto these basis images, we can effectively extract and eliminate the LDIV from the I-image without discarding the discriminant information. Experimental results confirm that FR with the proposed method outperforms that with existing IN methods under varying lighting conditions.

Original languageEnglish
Pages (from-to)348-358
Number of pages11
JournalSignal Processing
Volume120
DOIs
Publication statusPublished - 2016 Mar 1

Fingerprint

Face recognition
Lighting
Singular value decomposition

Keywords

  • Face recognition
  • Face relighting
  • Face restoration
  • Illumination normalization
  • Illumination subspace

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Control and Systems Engineering
  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition

Cite this

Retinex-based illumination normalization using class-based illumination subspace for robust face recognition. / Kim, Seung Wook; Jung, June Young; Yoo, Cheol Hwan; Ko, Sung-Jea.

In: Signal Processing, Vol. 120, 01.03.2016, p. 348-358.

Research output: Contribution to journalArticle

@article{6c920f845b71438f9b5a04eca8360470,
title = "Retinex-based illumination normalization using class-based illumination subspace for robust face recognition",
abstract = "Recent illumination normalization (IN) methods first decompose a face image into a reflectance (R)-image having a lighting-invariant characteristic and an illuminance (I)-image including shading and shadowing effects. An illumination-normalized I-image is then obtained by eliminating the lighting-dependent image variations (LDIV) from the I-image. Finally, the normalized I-and R-images are recombined for face recognition (FR). However, the decomposed-reflectance is often contaminated with the lighting effects. Moreover, the lighting normalization tends to remove the valuable discriminant information in the I-image. To address these problems, we employ the local edge-preserving filter to generate the R-image whereby the lighting-invariant information is well preserved. In addition, we propose a subspace-based IN method that can retain the large facial-structure in the I-image. To construct the proposed subspace, we calculate the LDIV within the same class of people from the training database of face images. Then, we apply the singular value decomposition to the calculated LDIV to obtain the basis images of the subspace. By projecting the I-image onto these basis images, we can effectively extract and eliminate the LDIV from the I-image without discarding the discriminant information. Experimental results confirm that FR with the proposed method outperforms that with existing IN methods under varying lighting conditions.",
keywords = "Face recognition, Face relighting, Face restoration, Illumination normalization, Illumination subspace",
author = "Kim, {Seung Wook} and Jung, {June Young} and Yoo, {Cheol Hwan} and Sung-Jea Ko",
year = "2016",
month = "3",
day = "1",
doi = "10.1016/j.sigpro.2015.09.028",
language = "English",
volume = "120",
pages = "348--358",
journal = "Signal Processing",
issn = "0165-1684",
publisher = "Elsevier",

}

TY - JOUR

T1 - Retinex-based illumination normalization using class-based illumination subspace for robust face recognition

AU - Kim, Seung Wook

AU - Jung, June Young

AU - Yoo, Cheol Hwan

AU - Ko, Sung-Jea

PY - 2016/3/1

Y1 - 2016/3/1

N2 - Recent illumination normalization (IN) methods first decompose a face image into a reflectance (R)-image having a lighting-invariant characteristic and an illuminance (I)-image including shading and shadowing effects. An illumination-normalized I-image is then obtained by eliminating the lighting-dependent image variations (LDIV) from the I-image. Finally, the normalized I-and R-images are recombined for face recognition (FR). However, the decomposed-reflectance is often contaminated with the lighting effects. Moreover, the lighting normalization tends to remove the valuable discriminant information in the I-image. To address these problems, we employ the local edge-preserving filter to generate the R-image whereby the lighting-invariant information is well preserved. In addition, we propose a subspace-based IN method that can retain the large facial-structure in the I-image. To construct the proposed subspace, we calculate the LDIV within the same class of people from the training database of face images. Then, we apply the singular value decomposition to the calculated LDIV to obtain the basis images of the subspace. By projecting the I-image onto these basis images, we can effectively extract and eliminate the LDIV from the I-image without discarding the discriminant information. Experimental results confirm that FR with the proposed method outperforms that with existing IN methods under varying lighting conditions.

AB - Recent illumination normalization (IN) methods first decompose a face image into a reflectance (R)-image having a lighting-invariant characteristic and an illuminance (I)-image including shading and shadowing effects. An illumination-normalized I-image is then obtained by eliminating the lighting-dependent image variations (LDIV) from the I-image. Finally, the normalized I-and R-images are recombined for face recognition (FR). However, the decomposed-reflectance is often contaminated with the lighting effects. Moreover, the lighting normalization tends to remove the valuable discriminant information in the I-image. To address these problems, we employ the local edge-preserving filter to generate the R-image whereby the lighting-invariant information is well preserved. In addition, we propose a subspace-based IN method that can retain the large facial-structure in the I-image. To construct the proposed subspace, we calculate the LDIV within the same class of people from the training database of face images. Then, we apply the singular value decomposition to the calculated LDIV to obtain the basis images of the subspace. By projecting the I-image onto these basis images, we can effectively extract and eliminate the LDIV from the I-image without discarding the discriminant information. Experimental results confirm that FR with the proposed method outperforms that with existing IN methods under varying lighting conditions.

KW - Face recognition

KW - Face relighting

KW - Face restoration

KW - Illumination normalization

KW - Illumination subspace

UR - http://www.scopus.com/inward/record.url?scp=84945289829&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84945289829&partnerID=8YFLogxK

U2 - 10.1016/j.sigpro.2015.09.028

DO - 10.1016/j.sigpro.2015.09.028

M3 - Article

VL - 120

SP - 348

EP - 358

JO - Signal Processing

JF - Signal Processing

SN - 0165-1684

ER -