Maximum neighborhood margin criterion in face recognition

Pang Ying Han, Andrew Beng Jin Teoh

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Feature extraction is a data analysis technique devoted to removing redundancy and extracting the most discriminative information. In face recognition, feature extractors are normally plagued with small sample size problems, in which the total number of training images is much smaller than the image dimensionality. Recently, an optimized facial feature extractor, maximum marginal criterion (MMC), was proposed. MMC computes an optimized projection by solving the generalized eigenvalue problem in a standard form that is free from inverse matrix operation, and thus it does not suffer from the small sample size problem. However, MMC is essentially a linear projection technique that relies on facial image pixel intensity to compute within- and between-class scatters. The nonlinear nature of faces restricts the discrimination of MMC. Hence, we propose an improved MMC, namely maximum neighborhood margin criterion (MNMC). Unlike MMC, which preserves global geometric structures that do not perfectly describe the underlying face manifold, MNMC seeks a projection that preserves local geometric structures via neighborhood preservation. This objective function leads to the enhancement of classification capability, and this is testified by experimental results. MNMC shows its performance superiority compared to MMC, especially in pose, illumination, and expression (PIE) and face recognition grand challenge (FRGC) databases.

Original languageEnglish
Article number047205
JournalOptical Engineering
Volume48
Issue number4
DOIs
Publication statusPublished - 2009 Dec 1

Fingerprint

Face recognition
margins
Redundancy
Feature extraction
Lighting
Pixels
projection
redundancy
pattern recognition
discrimination
education
eigenvalues
illumination
pixels
augmentation
matrices

All Science Journal Classification (ASJC) codes

  • Atomic and Molecular Physics, and Optics
  • Engineering(all)

Cite this

@article{6b3f5a0b03b7456384d75b6786362e99,
title = "Maximum neighborhood margin criterion in face recognition",
abstract = "Feature extraction is a data analysis technique devoted to removing redundancy and extracting the most discriminative information. In face recognition, feature extractors are normally plagued with small sample size problems, in which the total number of training images is much smaller than the image dimensionality. Recently, an optimized facial feature extractor, maximum marginal criterion (MMC), was proposed. MMC computes an optimized projection by solving the generalized eigenvalue problem in a standard form that is free from inverse matrix operation, and thus it does not suffer from the small sample size problem. However, MMC is essentially a linear projection technique that relies on facial image pixel intensity to compute within- and between-class scatters. The nonlinear nature of faces restricts the discrimination of MMC. Hence, we propose an improved MMC, namely maximum neighborhood margin criterion (MNMC). Unlike MMC, which preserves global geometric structures that do not perfectly describe the underlying face manifold, MNMC seeks a projection that preserves local geometric structures via neighborhood preservation. This objective function leads to the enhancement of classification capability, and this is testified by experimental results. MNMC shows its performance superiority compared to MMC, especially in pose, illumination, and expression (PIE) and face recognition grand challenge (FRGC) databases.",
author = "Han, {Pang Ying} and Teoh, {Andrew Beng Jin}",
year = "2009",
month = "12",
day = "1",
doi = "10.1117/1.3122033",
language = "English",
volume = "48",
journal = "Optical Engineering",
issn = "0091-3286",
publisher = "SPIE",
number = "4",

}

Maximum neighborhood margin criterion in face recognition. / Han, Pang Ying; Teoh, Andrew Beng Jin.

In: Optical Engineering, Vol. 48, No. 4, 047205, 01.12.2009.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Maximum neighborhood margin criterion in face recognition

AU - Han, Pang Ying

AU - Teoh, Andrew Beng Jin

PY - 2009/12/1

Y1 - 2009/12/1

N2 - Feature extraction is a data analysis technique devoted to removing redundancy and extracting the most discriminative information. In face recognition, feature extractors are normally plagued with small sample size problems, in which the total number of training images is much smaller than the image dimensionality. Recently, an optimized facial feature extractor, maximum marginal criterion (MMC), was proposed. MMC computes an optimized projection by solving the generalized eigenvalue problem in a standard form that is free from inverse matrix operation, and thus it does not suffer from the small sample size problem. However, MMC is essentially a linear projection technique that relies on facial image pixel intensity to compute within- and between-class scatters. The nonlinear nature of faces restricts the discrimination of MMC. Hence, we propose an improved MMC, namely maximum neighborhood margin criterion (MNMC). Unlike MMC, which preserves global geometric structures that do not perfectly describe the underlying face manifold, MNMC seeks a projection that preserves local geometric structures via neighborhood preservation. This objective function leads to the enhancement of classification capability, and this is testified by experimental results. MNMC shows its performance superiority compared to MMC, especially in pose, illumination, and expression (PIE) and face recognition grand challenge (FRGC) databases.

AB - Feature extraction is a data analysis technique devoted to removing redundancy and extracting the most discriminative information. In face recognition, feature extractors are normally plagued with small sample size problems, in which the total number of training images is much smaller than the image dimensionality. Recently, an optimized facial feature extractor, maximum marginal criterion (MMC), was proposed. MMC computes an optimized projection by solving the generalized eigenvalue problem in a standard form that is free from inverse matrix operation, and thus it does not suffer from the small sample size problem. However, MMC is essentially a linear projection technique that relies on facial image pixel intensity to compute within- and between-class scatters. The nonlinear nature of faces restricts the discrimination of MMC. Hence, we propose an improved MMC, namely maximum neighborhood margin criterion (MNMC). Unlike MMC, which preserves global geometric structures that do not perfectly describe the underlying face manifold, MNMC seeks a projection that preserves local geometric structures via neighborhood preservation. This objective function leads to the enhancement of classification capability, and this is testified by experimental results. MNMC shows its performance superiority compared to MMC, especially in pose, illumination, and expression (PIE) and face recognition grand challenge (FRGC) databases.

UR - http://www.scopus.com/inward/record.url?scp=77955257633&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77955257633&partnerID=8YFLogxK

U2 - 10.1117/1.3122033

DO - 10.1117/1.3122033

M3 - Article

VL - 48

JO - Optical Engineering

JF - Optical Engineering

SN - 0091-3286

IS - 4

M1 - 047205

ER -