The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition

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

8 Downloads (Pure)

Abstract

This paper presents the analysis of PCA-LDA behavior for face recognition using Singular Value Decomposition (SVD). The experimental results is shown to analyze face recognition performance, i.e. the impact of number of subjects, images per subject, training set size, and trade-off between the number of subjects and the number of images per subject on recognition performance, in relation with the number of PCA-LDA coefficients. The comparison of three classifiers, i.e. Euclidean Distance, Cosine Similarity, and Likelihood Ratio, are presented to obtain knowledge about their characteristics. All experimental evaluations are in the verification context. Based on the experimental results, the larger number of subjects and images per subject produced the better recognition performance. Regarding the number of subjects and images per subject trade-off, its indicated both of them influence the recognition performance. Otherwise, the image size also affect to recognition performance. PCA-LDA can perform low resolution image well up to 15x15 pixels and breaks down afterward. Regarding the p and ` coefficients, PCA-LDA has different behavior for each classifier.
Original languageEnglish
Title of host publicationProceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux
Subtitle of host publicationMay 31-1 June, 2018, University of Twente, Enschede, The Netherlands
EditorsLuuk Spreeuwers, Jasper Goseling
Place of PublicationEnschede
PublisherWerkgemeenschap voor Informatie- en Communicatietheorie (WIC)
Pages133-148
Number of pages16
ISBN (Print)978-90-365-4570-9
Publication statusPublished - 31 May 2018
Event39th Symposium on Information Theory and Signal Processing in the Benelux 2018 - University of Twente, Enschede, Netherlands
Duration: 31 May 20181 Jun 2018
Conference number: 39
https://www.utwente.nl/en/eemcs/sitb2018/

Conference

Conference39th Symposium on Information Theory and Signal Processing in the Benelux 2018
Abbreviated titleSITB
CountryNetherlands
CityEnschede
Period31/05/181/06/18
Internet address

Fingerprint

Discriminant analysis
Face recognition
Principal component analysis
Singular value decomposition
Image resolution
Pixels

Keywords

  • Face recognition
  • Principal component analysis
  • Linear discriminant analysis

Cite this

Lestriandoko, N. H., Spreeuwers, L., & Veldhuis, R. (2018). The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition. In L. Spreeuwers, & J. Goseling (Eds.), Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux: May 31-1 June, 2018, University of Twente, Enschede, The Netherlands (pp. 133-148). Enschede: Werkgemeenschap voor Informatie- en Communicatietheorie (WIC).
Lestriandoko, Nova Hadi ; Spreeuwers, Luuk ; Veldhuis, Raymond. / The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition. Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux: May 31-1 June, 2018, University of Twente, Enschede, The Netherlands. editor / Luuk Spreeuwers ; Jasper Goseling. Enschede : Werkgemeenschap voor Informatie- en Communicatietheorie (WIC), 2018. pp. 133-148
@inproceedings{767050bfb1774e8aa9c4dc701478a173,
title = "The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition",
abstract = "This paper presents the analysis of PCA-LDA behavior for face recognition using Singular Value Decomposition (SVD). The experimental results is shown to analyze face recognition performance, i.e. the impact of number of subjects, images per subject, training set size, and trade-off between the number of subjects and the number of images per subject on recognition performance, in relation with the number of PCA-LDA coefficients. The comparison of three classifiers, i.e. Euclidean Distance, Cosine Similarity, and Likelihood Ratio, are presented to obtain knowledge about their characteristics. All experimental evaluations are in the verification context. Based on the experimental results, the larger number of subjects and images per subject produced the better recognition performance. Regarding the number of subjects and images per subject trade-off, its indicated both of them influence the recognition performance. Otherwise, the image size also affect to recognition performance. PCA-LDA can perform low resolution image well up to 15x15 pixels and breaks down afterward. Regarding the p and ` coefficients, PCA-LDA has different behavior for each classifier.",
keywords = "Face recognition, Principal component analysis, Linear discriminant analysis",
author = "Lestriandoko, {Nova Hadi} and Luuk Spreeuwers and Raymond Veldhuis",
year = "2018",
month = "5",
day = "31",
language = "English",
isbn = "978-90-365-4570-9",
pages = "133--148",
editor = "Luuk Spreeuwers and Jasper Goseling",
booktitle = "Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux",
publisher = "Werkgemeenschap voor Informatie- en Communicatietheorie (WIC)",
address = "Netherlands",

}

Lestriandoko, NH, Spreeuwers, L & Veldhuis, R 2018, The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition. in L Spreeuwers & J Goseling (eds), Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux: May 31-1 June, 2018, University of Twente, Enschede, The Netherlands. Werkgemeenschap voor Informatie- en Communicatietheorie (WIC), Enschede, pp. 133-148, 39th Symposium on Information Theory and Signal Processing in the Benelux 2018, Enschede, Netherlands, 31/05/18.

The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition. / Lestriandoko, Nova Hadi ; Spreeuwers, Luuk; Veldhuis, Raymond.

Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux: May 31-1 June, 2018, University of Twente, Enschede, The Netherlands. ed. / Luuk Spreeuwers; Jasper Goseling. Enschede : Werkgemeenschap voor Informatie- en Communicatietheorie (WIC), 2018. p. 133-148.

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

TY - GEN

T1 - The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition

AU - Lestriandoko, Nova Hadi

AU - Spreeuwers, Luuk

AU - Veldhuis, Raymond

PY - 2018/5/31

Y1 - 2018/5/31

N2 - This paper presents the analysis of PCA-LDA behavior for face recognition using Singular Value Decomposition (SVD). The experimental results is shown to analyze face recognition performance, i.e. the impact of number of subjects, images per subject, training set size, and trade-off between the number of subjects and the number of images per subject on recognition performance, in relation with the number of PCA-LDA coefficients. The comparison of three classifiers, i.e. Euclidean Distance, Cosine Similarity, and Likelihood Ratio, are presented to obtain knowledge about their characteristics. All experimental evaluations are in the verification context. Based on the experimental results, the larger number of subjects and images per subject produced the better recognition performance. Regarding the number of subjects and images per subject trade-off, its indicated both of them influence the recognition performance. Otherwise, the image size also affect to recognition performance. PCA-LDA can perform low resolution image well up to 15x15 pixels and breaks down afterward. Regarding the p and ` coefficients, PCA-LDA has different behavior for each classifier.

AB - This paper presents the analysis of PCA-LDA behavior for face recognition using Singular Value Decomposition (SVD). The experimental results is shown to analyze face recognition performance, i.e. the impact of number of subjects, images per subject, training set size, and trade-off between the number of subjects and the number of images per subject on recognition performance, in relation with the number of PCA-LDA coefficients. The comparison of three classifiers, i.e. Euclidean Distance, Cosine Similarity, and Likelihood Ratio, are presented to obtain knowledge about their characteristics. All experimental evaluations are in the verification context. Based on the experimental results, the larger number of subjects and images per subject produced the better recognition performance. Regarding the number of subjects and images per subject trade-off, its indicated both of them influence the recognition performance. Otherwise, the image size also affect to recognition performance. PCA-LDA can perform low resolution image well up to 15x15 pixels and breaks down afterward. Regarding the p and ` coefficients, PCA-LDA has different behavior for each classifier.

KW - Face recognition

KW - Principal component analysis

KW - Linear discriminant analysis

M3 - Conference contribution

SN - 978-90-365-4570-9

SP - 133

EP - 148

BT - Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux

A2 - Spreeuwers, Luuk

A2 - Goseling, Jasper

PB - Werkgemeenschap voor Informatie- en Communicatietheorie (WIC)

CY - Enschede

ER -

Lestriandoko NH, Spreeuwers L, Veldhuis R. The Behavior of Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) for Face Recognition. In Spreeuwers L, Goseling J, editors, Proceedings of the 2018 Symposium on Information Theory and Signal Processing in the Benelux: May 31-1 June, 2018, University of Twente, Enschede, The Netherlands. Enschede: Werkgemeenschap voor Informatie- en Communicatietheorie (WIC). 2018. p. 133-148