On the effects of illumination normalization with LBP-based watchlist screening

Ibtihel Amara, Eric Granger, Abdenour Hadid

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

1 Citation (Scopus)

Abstract

Still-to-video face recognition (FR) is an important function in several video surveillance applications like watchlist screening, where faces captured over a network of video cameras are matched against reference stills belonging to target individuals. Screening of faces against a watchlist is a challenging problem due to variations in capturing conditions (e.g., pose and illumination), to camera inter-operability, and to the limited number of reference stills. In holistic approaches to FR, Local Binary Pattern (LBP) descriptors are often considered to represent facial captures and reference stills. Despite their efficiency, LBP descriptors are known as being sensitive to illumination changes. In this paper, the performance of still-to-video FR is compared when different passive illumination normalization techniques are applied prior to LBP feature extraction. This study focuses on representative retinex, self-quotient, diffusion, filtering, means de-noising, retina, wavelet and frequency-based techniques that are suitable for fast and accurate face screening. Experimental results obtained with videos from the Chokepoint dataset indicate that, although Multi-Scale Weberfaces and Tan and Triggs techniques tend to outperform others, the benefits of these techniques varies considerably according to the individual and illumination conditions. Results suggest that a combination of these techniques should be selected dynamically based on changing capture conditions.

Original languageEnglish
Title of host publicationComputer Vision - ECCV 2014 Workshops, Proceedings
PublisherSpringer Verlag
Pages173-188
Number of pages16
ISBN (Electronic)9783319161808
DOIs
Publication statusPublished - 1 Jan 2015
Externally publishedYes
Event13th European Conference on Computer Vision, ECCV 2014 - Zurich, Switzerland
Duration: 6 Sep 201412 Sep 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8926
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other13th European Conference on Computer Vision, ECCV 2014
CountrySwitzerland
CityZurich
Period6/9/1412/9/14

Fingerprint

Normalization
Screening
Illumination
Face recognition
Lighting
Binary
Face Recognition
Face
Descriptors
Camera
Video cameras
Interoperability
Retina
Video Surveillance
Feature extraction
Denoising
Cameras
Feature Extraction
Quotient
Wavelets

Keywords

  • Face screening
  • Illumination normalization
  • Local binary patterns
  • Still-to-video face recognition
  • Video surveillance

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Amara, I., Granger, E., & Hadid, A. (2015). On the effects of illumination normalization with LBP-based watchlist screening. In Computer Vision - ECCV 2014 Workshops, Proceedings (pp. 173-188). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8926). Springer Verlag. https://doi.org/10.1007/978-3-319-16181-5_13

On the effects of illumination normalization with LBP-based watchlist screening. / Amara, Ibtihel; Granger, Eric; Hadid, Abdenour.

Computer Vision - ECCV 2014 Workshops, Proceedings. Springer Verlag, 2015. p. 173-188 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8926).

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

Amara, I, Granger, E & Hadid, A 2015, On the effects of illumination normalization with LBP-based watchlist screening. in Computer Vision - ECCV 2014 Workshops, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8926, Springer Verlag, pp. 173-188, 13th European Conference on Computer Vision, ECCV 2014, Zurich, Switzerland, 6/9/14. https://doi.org/10.1007/978-3-319-16181-5_13
Amara I, Granger E, Hadid A. On the effects of illumination normalization with LBP-based watchlist screening. In Computer Vision - ECCV 2014 Workshops, Proceedings. Springer Verlag. 2015. p. 173-188. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16181-5_13
Amara, Ibtihel ; Granger, Eric ; Hadid, Abdenour. / On the effects of illumination normalization with LBP-based watchlist screening. Computer Vision - ECCV 2014 Workshops, Proceedings. Springer Verlag, 2015. pp. 173-188 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{fae09b02f1a048dbb335ee0b80758262,
title = "On the effects of illumination normalization with LBP-based watchlist screening",
abstract = "Still-to-video face recognition (FR) is an important function in several video surveillance applications like watchlist screening, where faces captured over a network of video cameras are matched against reference stills belonging to target individuals. Screening of faces against a watchlist is a challenging problem due to variations in capturing conditions (e.g., pose and illumination), to camera inter-operability, and to the limited number of reference stills. In holistic approaches to FR, Local Binary Pattern (LBP) descriptors are often considered to represent facial captures and reference stills. Despite their efficiency, LBP descriptors are known as being sensitive to illumination changes. In this paper, the performance of still-to-video FR is compared when different passive illumination normalization techniques are applied prior to LBP feature extraction. This study focuses on representative retinex, self-quotient, diffusion, filtering, means de-noising, retina, wavelet and frequency-based techniques that are suitable for fast and accurate face screening. Experimental results obtained with videos from the Chokepoint dataset indicate that, although Multi-Scale Weberfaces and Tan and Triggs techniques tend to outperform others, the benefits of these techniques varies considerably according to the individual and illumination conditions. Results suggest that a combination of these techniques should be selected dynamically based on changing capture conditions.",
keywords = "Face screening, Illumination normalization, Local binary patterns, Still-to-video face recognition, Video surveillance",
author = "Ibtihel Amara and Eric Granger and Abdenour Hadid",
year = "2015",
month = "1",
day = "1",
doi = "10.1007/978-3-319-16181-5_13",
language = "English",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "173--188",
booktitle = "Computer Vision - ECCV 2014 Workshops, Proceedings",

}

TY - GEN

T1 - On the effects of illumination normalization with LBP-based watchlist screening

AU - Amara, Ibtihel

AU - Granger, Eric

AU - Hadid, Abdenour

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Still-to-video face recognition (FR) is an important function in several video surveillance applications like watchlist screening, where faces captured over a network of video cameras are matched against reference stills belonging to target individuals. Screening of faces against a watchlist is a challenging problem due to variations in capturing conditions (e.g., pose and illumination), to camera inter-operability, and to the limited number of reference stills. In holistic approaches to FR, Local Binary Pattern (LBP) descriptors are often considered to represent facial captures and reference stills. Despite their efficiency, LBP descriptors are known as being sensitive to illumination changes. In this paper, the performance of still-to-video FR is compared when different passive illumination normalization techniques are applied prior to LBP feature extraction. This study focuses on representative retinex, self-quotient, diffusion, filtering, means de-noising, retina, wavelet and frequency-based techniques that are suitable for fast and accurate face screening. Experimental results obtained with videos from the Chokepoint dataset indicate that, although Multi-Scale Weberfaces and Tan and Triggs techniques tend to outperform others, the benefits of these techniques varies considerably according to the individual and illumination conditions. Results suggest that a combination of these techniques should be selected dynamically based on changing capture conditions.

AB - Still-to-video face recognition (FR) is an important function in several video surveillance applications like watchlist screening, where faces captured over a network of video cameras are matched against reference stills belonging to target individuals. Screening of faces against a watchlist is a challenging problem due to variations in capturing conditions (e.g., pose and illumination), to camera inter-operability, and to the limited number of reference stills. In holistic approaches to FR, Local Binary Pattern (LBP) descriptors are often considered to represent facial captures and reference stills. Despite their efficiency, LBP descriptors are known as being sensitive to illumination changes. In this paper, the performance of still-to-video FR is compared when different passive illumination normalization techniques are applied prior to LBP feature extraction. This study focuses on representative retinex, self-quotient, diffusion, filtering, means de-noising, retina, wavelet and frequency-based techniques that are suitable for fast and accurate face screening. Experimental results obtained with videos from the Chokepoint dataset indicate that, although Multi-Scale Weberfaces and Tan and Triggs techniques tend to outperform others, the benefits of these techniques varies considerably according to the individual and illumination conditions. Results suggest that a combination of these techniques should be selected dynamically based on changing capture conditions.

KW - Face screening

KW - Illumination normalization

KW - Local binary patterns

KW - Still-to-video face recognition

KW - Video surveillance

UR - http://www.scopus.com/inward/record.url?scp=84928817514&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84928817514&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-16181-5_13

DO - 10.1007/978-3-319-16181-5_13

M3 - Conference contribution

AN - SCOPUS:84928817514

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 173

EP - 188

BT - Computer Vision - ECCV 2014 Workshops, Proceedings

PB - Springer Verlag

ER -