Robust 3D Face Recognition in the Presence of Realistic Occlusions

Nese Alyuz, B. Gökberk, Lieuwe Jan Spreeuwers, Raymond N.J. Veldhuis, Lale Akarun

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

19 Citations (Scopus)

Abstract

Facial occlusions pose significant problems for automatic face recognition systems. In this work, we propose a novel occlusion-resistant three-dimensional (3D) facial identification system. We show that, under extreme occlusions due to hair, hands, and eyeglasses, typical 3D face recognition systems exhibit poor performance. In order to deal with occlusions, our proposed system employs occlusion-resistant registration, occlusion detection, and regional classifiers. A two-step registration module first detects the nose region on the curvedness-weighted convex shape index map, and then performs fine alignment using nose-based Iterative Closest Point (ICP) algorithm. Occluded areas are determined automatically via a generic face model. After non-facial parts introduced by occlusions are removed, a variant of Gappy Principal Component Analysis (Gappy PCA) is used to restore the full face from occlusion-free facial surfaces. Experimental results obtained on realistically occluded facial images from the Bosphorus 3D face database shows that, with the use of score-level fusion of regional Linear Discriminant Analysis (LDA) classifiers, the proposed method improves rank-1 identification accuracy significantly: from 76.12% to 94.23%.
Original languageUndefined
Title of host publication5th IAPR International Conference on Biometrics, ICB 2012
Place of PublicationUSA
PublisherIEEE Computer Society
Pages111-118
Number of pages8
ISBN (Print)978-1-4673-0396-5
DOIs
Publication statusPublished - 1 Apr 2012
Event5th IAPR International Conference on Biometrics, ICB 2012 - ITC Sheraton, New Delhi, India
Duration: 29 Mar 20121 Apr 2012
http://icb12.iiitd.ac.in/

Publication series

Name
PublisherIEEE Computer Society

Conference

Conference5th IAPR International Conference on Biometrics, ICB 2012
Abbreviated titleICB
CountryIndia
CityNew Delhi
Period29/03/121/04/12
Internet address

Keywords

  • METIS-287860
  • Biometrics
  • IR-81247
  • Pattern Recognition
  • EWI-21548
  • Computer Vision
  • SCS-Safety
  • CR-I.5
  • Face Recognition

Cite this

Alyuz, N., Gökberk, B., Spreeuwers, L. J., Veldhuis, R. N. J., & Akarun, L. (2012). Robust 3D Face Recognition in the Presence of Realistic Occlusions. In 5th IAPR International Conference on Biometrics, ICB 2012 (pp. 111-118). USA: IEEE Computer Society. https://doi.org/10.1109/ICB.2012.6199767
Alyuz, Nese ; Gökberk, B. ; Spreeuwers, Lieuwe Jan ; Veldhuis, Raymond N.J. ; Akarun, Lale. / Robust 3D Face Recognition in the Presence of Realistic Occlusions. 5th IAPR International Conference on Biometrics, ICB 2012. USA : IEEE Computer Society, 2012. pp. 111-118
@inproceedings{a8baac8854a04d4da2848bc8eeedc6a3,
title = "Robust 3D Face Recognition in the Presence of Realistic Occlusions",
abstract = "Facial occlusions pose significant problems for automatic face recognition systems. In this work, we propose a novel occlusion-resistant three-dimensional (3D) facial identification system. We show that, under extreme occlusions due to hair, hands, and eyeglasses, typical 3D face recognition systems exhibit poor performance. In order to deal with occlusions, our proposed system employs occlusion-resistant registration, occlusion detection, and regional classifiers. A two-step registration module first detects the nose region on the curvedness-weighted convex shape index map, and then performs fine alignment using nose-based Iterative Closest Point (ICP) algorithm. Occluded areas are determined automatically via a generic face model. After non-facial parts introduced by occlusions are removed, a variant of Gappy Principal Component Analysis (Gappy PCA) is used to restore the full face from occlusion-free facial surfaces. Experimental results obtained on realistically occluded facial images from the Bosphorus 3D face database shows that, with the use of score-level fusion of regional Linear Discriminant Analysis (LDA) classifiers, the proposed method improves rank-1 identification accuracy significantly: from 76.12{\%} to 94.23{\%}.",
keywords = "METIS-287860, Biometrics, IR-81247, Pattern Recognition, EWI-21548, Computer Vision, SCS-Safety, CR-I.5, Face Recognition",
author = "Nese Alyuz and B. G{\"o}kberk and Spreeuwers, {Lieuwe Jan} and Veldhuis, {Raymond N.J.} and Lale Akarun",
note = "10.1109/ICB.2012.6199767",
year = "2012",
month = "4",
day = "1",
doi = "10.1109/ICB.2012.6199767",
language = "Undefined",
isbn = "978-1-4673-0396-5",
publisher = "IEEE Computer Society",
pages = "111--118",
booktitle = "5th IAPR International Conference on Biometrics, ICB 2012",
address = "United States",

}

Alyuz, N, Gökberk, B, Spreeuwers, LJ, Veldhuis, RNJ & Akarun, L 2012, Robust 3D Face Recognition in the Presence of Realistic Occlusions. in 5th IAPR International Conference on Biometrics, ICB 2012. IEEE Computer Society, USA, pp. 111-118, 5th IAPR International Conference on Biometrics, ICB 2012, New Delhi, India, 29/03/12. https://doi.org/10.1109/ICB.2012.6199767

Robust 3D Face Recognition in the Presence of Realistic Occlusions. / Alyuz, Nese; Gökberk, B.; Spreeuwers, Lieuwe Jan; Veldhuis, Raymond N.J.; Akarun, Lale.

5th IAPR International Conference on Biometrics, ICB 2012. USA : IEEE Computer Society, 2012. p. 111-118.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Robust 3D Face Recognition in the Presence of Realistic Occlusions

AU - Alyuz, Nese

AU - Gökberk, B.

AU - Spreeuwers, Lieuwe Jan

AU - Veldhuis, Raymond N.J.

AU - Akarun, Lale

N1 - 10.1109/ICB.2012.6199767

PY - 2012/4/1

Y1 - 2012/4/1

N2 - Facial occlusions pose significant problems for automatic face recognition systems. In this work, we propose a novel occlusion-resistant three-dimensional (3D) facial identification system. We show that, under extreme occlusions due to hair, hands, and eyeglasses, typical 3D face recognition systems exhibit poor performance. In order to deal with occlusions, our proposed system employs occlusion-resistant registration, occlusion detection, and regional classifiers. A two-step registration module first detects the nose region on the curvedness-weighted convex shape index map, and then performs fine alignment using nose-based Iterative Closest Point (ICP) algorithm. Occluded areas are determined automatically via a generic face model. After non-facial parts introduced by occlusions are removed, a variant of Gappy Principal Component Analysis (Gappy PCA) is used to restore the full face from occlusion-free facial surfaces. Experimental results obtained on realistically occluded facial images from the Bosphorus 3D face database shows that, with the use of score-level fusion of regional Linear Discriminant Analysis (LDA) classifiers, the proposed method improves rank-1 identification accuracy significantly: from 76.12% to 94.23%.

AB - Facial occlusions pose significant problems for automatic face recognition systems. In this work, we propose a novel occlusion-resistant three-dimensional (3D) facial identification system. We show that, under extreme occlusions due to hair, hands, and eyeglasses, typical 3D face recognition systems exhibit poor performance. In order to deal with occlusions, our proposed system employs occlusion-resistant registration, occlusion detection, and regional classifiers. A two-step registration module first detects the nose region on the curvedness-weighted convex shape index map, and then performs fine alignment using nose-based Iterative Closest Point (ICP) algorithm. Occluded areas are determined automatically via a generic face model. After non-facial parts introduced by occlusions are removed, a variant of Gappy Principal Component Analysis (Gappy PCA) is used to restore the full face from occlusion-free facial surfaces. Experimental results obtained on realistically occluded facial images from the Bosphorus 3D face database shows that, with the use of score-level fusion of regional Linear Discriminant Analysis (LDA) classifiers, the proposed method improves rank-1 identification accuracy significantly: from 76.12% to 94.23%.

KW - METIS-287860

KW - Biometrics

KW - IR-81247

KW - Pattern Recognition

KW - EWI-21548

KW - Computer Vision

KW - SCS-Safety

KW - CR-I.5

KW - Face Recognition

U2 - 10.1109/ICB.2012.6199767

DO - 10.1109/ICB.2012.6199767

M3 - Conference contribution

SN - 978-1-4673-0396-5

SP - 111

EP - 118

BT - 5th IAPR International Conference on Biometrics, ICB 2012

PB - IEEE Computer Society

CY - USA

ER -

Alyuz N, Gökberk B, Spreeuwers LJ, Veldhuis RNJ, Akarun L. Robust 3D Face Recognition in the Presence of Realistic Occlusions. In 5th IAPR International Conference on Biometrics, ICB 2012. USA: IEEE Computer Society. 2012. p. 111-118 https://doi.org/10.1109/ICB.2012.6199767