Principal component analysis of image gradient orientations for face recognition

Georgios Tzimiropoulos; Stefanos Zafeiriou; Maja Pantic

doi:10.1109/FG.2011.5771457

Principal component analysis of image gradient orientations for face recognition

Georgios Tzimiropoulos, Stefanos Zafeiriou, Maja Pantic

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

34 Citations (Scopus)

1 Downloads (Pure)

Abstract

We introduce the notion of Principal Component Analysis (PCA) of image gradient orientations. As image data is typically noisy, but noise is substantially different from Gaussian, traditional PCA of pixel intensities very often fails to estimate reliably the low-dimensional subspace of a given data population. We show that replacing intensities with gradient orientations and the _2 norm with a cosine-based distance measure offers, to some extend, a remedy to this problem. Our scheme requires the eigen-decomposition of a covariance matrix and is as computationally efficient as standard _2 intensitybased PCA. We demonstrate some of its favorable properties for the application of face recognition.

Original language	Undefined
Title of host publication	Proceedings of the IEEE International Conference on Automatic Face & Gesture Recognition and Workshops (FG 2011)
Place of Publication	USA
Publisher	IEEE
Pages	553-558
Number of pages	6
ISBN (Print)	978-1-4244-9140-7
DOIs	https://doi.org/10.1109/FG.2011.5771457
Publication status	Published - Mar 2011
Event	9th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2011 - Santa Barbara, United States Duration: 21 Mar 2011 → 25 Mar 2011 Conference number: 9

Publication series

Name
Publisher	IEEE Computer Society

Conference

Conference	9th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2011
Abbreviated title	FG
Country/Territory	United States
City	Santa Barbara
Period	21/03/11 → 25/03/11

Keywords

METIS-285026
IR-79432
Generators
Image reconstruction
Robustness
HMI-MI: MULTIMODAL INTERACTIONS
Face
Principal component analysis
Pixel
EWI-21324
EC Grant Agreement nr.: FP7/231287
Lighting

Access to Document

10.1109/FG.2011.5771457

http://eprints.eemcs.utwente.nl/secure2/21324/01/pantic_principal_component_analysis_of.pdf

Cite this

@inproceedings{bc4e30a8f7274b00bc30dd2db89b4ed9,

title = "Principal component analysis of image gradient orientations for face recognition",

abstract = "We introduce the notion of Principal Component Analysis (PCA) of image gradient orientations. As image data is typically noisy, but noise is substantially different from Gaussian, traditional PCA of pixel intensities very often fails to estimate reliably the low-dimensional subspace of a given data population. We show that replacing intensities with gradient orientations and the _2 norm with a cosine-based distance measure offers, to some extend, a remedy to this problem. Our scheme requires the eigen-decomposition of a covariance matrix and is as computationally efficient as standard _2 intensitybased PCA. We demonstrate some of its favorable properties for the application of face recognition.",

keywords = "METIS-285026, IR-79432, Generators, Image reconstruction, Robustness, HMI-MI: MULTIMODAL INTERACTIONS, Face, Principal component analysis, Pixel, EWI-21324, EC Grant Agreement nr.: FP7/231287, Lighting",

author = "Georgios Tzimiropoulos and Stefanos Zafeiriou and Maja Pantic",

note = "10.1109/FG.2011.5771457 ; 9th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2011 ; Conference date: 21-03-2011 Through 25-03-2011",

year = "2011",

month = mar,

doi = "10.1109/FG.2011.5771457",

language = "Undefined",

isbn = "978-1-4244-9140-7",

publisher = "IEEE",

pages = "553--558",

booktitle = "Proceedings of the IEEE International Conference on Automatic Face & Gesture Recognition and Workshops (FG 2011)",

address = "United States",

}

Tzimiropoulos, G, Zafeiriou, S & Pantic, M 2011, Principal component analysis of image gradient orientations for face recognition. in Proceedings of the IEEE International Conference on Automatic Face & Gesture Recognition and Workshops (FG 2011). IEEE, USA, pp. 553-558, 9th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2011, Santa Barbara, California, United States, 21/03/11. https://doi.org/10.1109/FG.2011.5771457

Principal component analysis of image gradient orientations for face recognition. / Tzimiropoulos, Georgios; Zafeiriou, Stefanos; Pantic, Maja.
Proceedings of the IEEE International Conference on Automatic Face & Gesture Recognition and Workshops (FG 2011). USA: IEEE, 2011. p. 553-558.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Principal component analysis of image gradient orientations for face recognition

AU - Tzimiropoulos, Georgios

AU - Zafeiriou, Stefanos

AU - Pantic, Maja

N1 - Conference code: 9

PY - 2011/3

Y1 - 2011/3

N2 - We introduce the notion of Principal Component Analysis (PCA) of image gradient orientations. As image data is typically noisy, but noise is substantially different from Gaussian, traditional PCA of pixel intensities very often fails to estimate reliably the low-dimensional subspace of a given data population. We show that replacing intensities with gradient orientations and the _2 norm with a cosine-based distance measure offers, to some extend, a remedy to this problem. Our scheme requires the eigen-decomposition of a covariance matrix and is as computationally efficient as standard _2 intensitybased PCA. We demonstrate some of its favorable properties for the application of face recognition.

AB - We introduce the notion of Principal Component Analysis (PCA) of image gradient orientations. As image data is typically noisy, but noise is substantially different from Gaussian, traditional PCA of pixel intensities very often fails to estimate reliably the low-dimensional subspace of a given data population. We show that replacing intensities with gradient orientations and the _2 norm with a cosine-based distance measure offers, to some extend, a remedy to this problem. Our scheme requires the eigen-decomposition of a covariance matrix and is as computationally efficient as standard _2 intensitybased PCA. We demonstrate some of its favorable properties for the application of face recognition.

KW - METIS-285026

KW - IR-79432

KW - Generators

KW - Image reconstruction

KW - Robustness

KW - HMI-MI: MULTIMODAL INTERACTIONS

KW - Face

KW - Principal component analysis

KW - Pixel

KW - EWI-21324

KW - EC Grant Agreement nr.: FP7/231287

KW - Lighting

U2 - 10.1109/FG.2011.5771457

DO - 10.1109/FG.2011.5771457

M3 - Conference contribution

SN - 978-1-4244-9140-7

SP - 553

EP - 558

BT - Proceedings of the IEEE International Conference on Automatic Face & Gesture Recognition and Workshops (FG 2011)

PB - IEEE

CY - USA

T2 - 9th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2011

Y2 - 21 March 2011 through 25 March 2011

ER -