2.5D face recognition under tensor manifold metrics

Lee Ying Chong, Beng Jin Teoh, Thian Song Ong, Siew Chin Chong

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

4 Citations (Scopus)

Abstract

Gabor-based region covariance matrix (GRCM) is a very flexible face descriptor where it allows different combination of features to be fused to construct a covariance matrix. GRCM resides on Tensor manifold where the computation of geodesic distance between two points requires the consideration of geometry characteristics of the manifold. Affine Invariant Riemannian Metric (AIRM) is the most widely used geodesic distance metric. However, it is computationally heavy. This paper investigates several geodesic distance metrics on Tensor manifold to find out the alternative speedy method for 2.5D face recognition using GRCM. Besides, we propose a feature-level fusion for 2.5D partial and 2D data to enhance the recognition performance.

Original languageEnglish
Title of host publicationNeural Information Processing - 21st International Conference, ICONIP 2014, Proceedings
EditorsChu Kiong Loo, Kok Wai Wong, Keem Siah Yap, Kaizhu Huang, Andrew Teoh
PublisherSpringer Verlag
Pages653-660
Number of pages8
ISBN (Electronic)9783319126425
Publication statusPublished - 2014 Jan 1

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8836
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Fingerprint

Face recognition
Covariance matrix
Face Recognition
Geodesic Distance
Tensors
Tensor
Metric
Distance Metric
Affine Invariant
Invariant Metric
Riemannian Metric
Descriptors
Fusion
Fusion reactions
Face
Partial
Geometry
Alternatives

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Chong, L. Y., Teoh, B. J., Ong, T. S., & Chong, S. C. (2014). 2.5D face recognition under tensor manifold metrics. In C. K. Loo, K. W. Wong, K. S. Yap, K. Huang, & A. Teoh (Eds.), Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings (pp. 653-660). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8836). Springer Verlag.
Chong, Lee Ying ; Teoh, Beng Jin ; Ong, Thian Song ; Chong, Siew Chin. / 2.5D face recognition under tensor manifold metrics. Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings. editor / Chu Kiong Loo ; Kok Wai Wong ; Keem Siah Yap ; Kaizhu Huang ; Andrew Teoh. Springer Verlag, 2014. pp. 653-660 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{529c37c2aad640d58f1fd52d0abb9360,
title = "2.5D face recognition under tensor manifold metrics",
abstract = "Gabor-based region covariance matrix (GRCM) is a very flexible face descriptor where it allows different combination of features to be fused to construct a covariance matrix. GRCM resides on Tensor manifold where the computation of geodesic distance between two points requires the consideration of geometry characteristics of the manifold. Affine Invariant Riemannian Metric (AIRM) is the most widely used geodesic distance metric. However, it is computationally heavy. This paper investigates several geodesic distance metrics on Tensor manifold to find out the alternative speedy method for 2.5D face recognition using GRCM. Besides, we propose a feature-level fusion for 2.5D partial and 2D data to enhance the recognition performance.",
author = "Chong, {Lee Ying} and Teoh, {Beng Jin} and Ong, {Thian Song} and Chong, {Siew Chin}",
year = "2014",
month = "1",
day = "1",
language = "English",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "653--660",
editor = "Loo, {Chu Kiong} and Wong, {Kok Wai} and Yap, {Keem Siah} and Kaizhu Huang and Andrew Teoh",
booktitle = "Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings",
address = "Germany",

}

Chong, LY, Teoh, BJ, Ong, TS & Chong, SC 2014, 2.5D face recognition under tensor manifold metrics. in CK Loo, KW Wong, KS Yap, K Huang & A Teoh (eds), Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8836, Springer Verlag, pp. 653-660.

2.5D face recognition under tensor manifold metrics. / Chong, Lee Ying; Teoh, Beng Jin; Ong, Thian Song; Chong, Siew Chin.

Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings. ed. / Chu Kiong Loo; Kok Wai Wong; Keem Siah Yap; Kaizhu Huang; Andrew Teoh. Springer Verlag, 2014. p. 653-660 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8836).

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

TY - GEN

T1 - 2.5D face recognition under tensor manifold metrics

AU - Chong, Lee Ying

AU - Teoh, Beng Jin

AU - Ong, Thian Song

AU - Chong, Siew Chin

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Gabor-based region covariance matrix (GRCM) is a very flexible face descriptor where it allows different combination of features to be fused to construct a covariance matrix. GRCM resides on Tensor manifold where the computation of geodesic distance between two points requires the consideration of geometry characteristics of the manifold. Affine Invariant Riemannian Metric (AIRM) is the most widely used geodesic distance metric. However, it is computationally heavy. This paper investigates several geodesic distance metrics on Tensor manifold to find out the alternative speedy method for 2.5D face recognition using GRCM. Besides, we propose a feature-level fusion for 2.5D partial and 2D data to enhance the recognition performance.

AB - Gabor-based region covariance matrix (GRCM) is a very flexible face descriptor where it allows different combination of features to be fused to construct a covariance matrix. GRCM resides on Tensor manifold where the computation of geodesic distance between two points requires the consideration of geometry characteristics of the manifold. Affine Invariant Riemannian Metric (AIRM) is the most widely used geodesic distance metric. However, it is computationally heavy. This paper investigates several geodesic distance metrics on Tensor manifold to find out the alternative speedy method for 2.5D face recognition using GRCM. Besides, we propose a feature-level fusion for 2.5D partial and 2D data to enhance the recognition performance.

UR - http://www.scopus.com/inward/record.url?scp=84909991525&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84909991525&partnerID=8YFLogxK

M3 - Conference contribution

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

SP - 653

EP - 660

BT - Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings

A2 - Loo, Chu Kiong

A2 - Wong, Kok Wai

A2 - Yap, Keem Siah

A2 - Huang, Kaizhu

A2 - Teoh, Andrew

PB - Springer Verlag

ER -

Chong LY, Teoh BJ, Ong TS, Chong SC. 2.5D face recognition under tensor manifold metrics. In Loo CK, Wong KW, Yap KS, Huang K, Teoh A, editors, Neural Information Processing - 21st International Conference, ICONIP 2014, Proceedings. Springer Verlag. 2014. p. 653-660. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).