Orthogonal filter banks with region Log-TiedRank covariance matrices for face recognition

Cong Jie Ng, Cheng Yaw Low, Kar Ann Toh, Jaihie Kim, Beng Jin Teoh

Research output: Contribution to journalArticle

Abstract

With the capability of fusing varying features from a specific image region, the Region Covariance Matrices (RCM) image descriptor has been evidenced plausible in face recognition. However, a systematic study for RCM, regarding which features to be fused in particular, remains absent. This paper therefore explores several features derived from the orthogonal filter ensembles, i.e., Identity Transform, Discrete Haar Transform, Discrete Cosine Transform, and Karhunen-Loève Transform, for feature encoding in RCM. Aside from that, we also outline a RCM variant, dubbed Region Log-TiedRank Covariance Matrices (RLTCM) in this paper. The RLTCM descriptor, on average, exhibits dramatic performance gain over RCM as well as state-of-the-art descriptors, especially when probe sets far deviated from the face gallery. Furthermore, we discern that the RLTCM descriptor defined based on Identity Transform, i.e., the simplest form of orthogonal filters, and other learning-free orthogonal filters yield impressive performance on par with the learning-based counterparts.

Original languageEnglish
Pages (from-to)548-560
Number of pages13
JournalJournal of Visual Communication and Image Representation
Volume55
DOIs
Publication statusPublished - 2018 Aug 1

Fingerprint

Filter banks
Face recognition
Covariance matrix
Mathematical transformations
Discrete cosine transforms

All Science Journal Classification (ASJC) codes

  • Signal Processing
  • Media Technology
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

Cite this

@article{fd3af2117c524839b4c81c52bd87358b,
title = "Orthogonal filter banks with region Log-TiedRank covariance matrices for face recognition",
abstract = "With the capability of fusing varying features from a specific image region, the Region Covariance Matrices (RCM) image descriptor has been evidenced plausible in face recognition. However, a systematic study for RCM, regarding which features to be fused in particular, remains absent. This paper therefore explores several features derived from the orthogonal filter ensembles, i.e., Identity Transform, Discrete Haar Transform, Discrete Cosine Transform, and Karhunen-Lo{\`e}ve Transform, for feature encoding in RCM. Aside from that, we also outline a RCM variant, dubbed Region Log-TiedRank Covariance Matrices (RLTCM) in this paper. The RLTCM descriptor, on average, exhibits dramatic performance gain over RCM as well as state-of-the-art descriptors, especially when probe sets far deviated from the face gallery. Furthermore, we discern that the RLTCM descriptor defined based on Identity Transform, i.e., the simplest form of orthogonal filters, and other learning-free orthogonal filters yield impressive performance on par with the learning-based counterparts.",
author = "Ng, {Cong Jie} and Low, {Cheng Yaw} and Toh, {Kar Ann} and Jaihie Kim and Teoh, {Beng Jin}",
year = "2018",
month = "8",
day = "1",
doi = "10.1016/j.jvcir.2018.07.002",
language = "English",
volume = "55",
pages = "548--560",
journal = "Journal of Visual Communication and Image Representation",
issn = "1047-3203",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Orthogonal filter banks with region Log-TiedRank covariance matrices for face recognition

AU - Ng, Cong Jie

AU - Low, Cheng Yaw

AU - Toh, Kar Ann

AU - Kim, Jaihie

AU - Teoh, Beng Jin

PY - 2018/8/1

Y1 - 2018/8/1

N2 - With the capability of fusing varying features from a specific image region, the Region Covariance Matrices (RCM) image descriptor has been evidenced plausible in face recognition. However, a systematic study for RCM, regarding which features to be fused in particular, remains absent. This paper therefore explores several features derived from the orthogonal filter ensembles, i.e., Identity Transform, Discrete Haar Transform, Discrete Cosine Transform, and Karhunen-Loève Transform, for feature encoding in RCM. Aside from that, we also outline a RCM variant, dubbed Region Log-TiedRank Covariance Matrices (RLTCM) in this paper. The RLTCM descriptor, on average, exhibits dramatic performance gain over RCM as well as state-of-the-art descriptors, especially when probe sets far deviated from the face gallery. Furthermore, we discern that the RLTCM descriptor defined based on Identity Transform, i.e., the simplest form of orthogonal filters, and other learning-free orthogonal filters yield impressive performance on par with the learning-based counterparts.

AB - With the capability of fusing varying features from a specific image region, the Region Covariance Matrices (RCM) image descriptor has been evidenced plausible in face recognition. However, a systematic study for RCM, regarding which features to be fused in particular, remains absent. This paper therefore explores several features derived from the orthogonal filter ensembles, i.e., Identity Transform, Discrete Haar Transform, Discrete Cosine Transform, and Karhunen-Loève Transform, for feature encoding in RCM. Aside from that, we also outline a RCM variant, dubbed Region Log-TiedRank Covariance Matrices (RLTCM) in this paper. The RLTCM descriptor, on average, exhibits dramatic performance gain over RCM as well as state-of-the-art descriptors, especially when probe sets far deviated from the face gallery. Furthermore, we discern that the RLTCM descriptor defined based on Identity Transform, i.e., the simplest form of orthogonal filters, and other learning-free orthogonal filters yield impressive performance on par with the learning-based counterparts.

UR - http://www.scopus.com/inward/record.url?scp=85049984371&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049984371&partnerID=8YFLogxK

U2 - 10.1016/j.jvcir.2018.07.002

DO - 10.1016/j.jvcir.2018.07.002

M3 - Article

VL - 55

SP - 548

EP - 560

JO - Journal of Visual Communication and Image Representation

JF - Journal of Visual Communication and Image Representation

SN - 1047-3203

ER -