Toward a practical face recognition system: Robust alignment and illumination by sparse representation

Andrew Wagner, John Wright, Arvind Ganesh, Zihan Zhou, Hossein Mobahi, Yi Ma

Research output: Contribution to journalArticle

437 Citations (Scopus)

Abstract

Many classic and contemporary face recognition algorithms work well on public data sets, but degrade sharply when they are used in a real recognition system. This is mostly due to the difficulty of simultaneously handling variations in illumination, image misalignment, and occlusion in the test image. We consider a scenario where the training images are well controlled and test images are only loosely controlled. We propose a conceptually simple face recognition system that achieves a high degree of robustness and stability to illumination variation, image misalignment, and partial occlusion. The system uses tools from sparse representation to align a test face image to a set of frontal training images. The region of attraction of our alignment algorithm is computed empirically for public face data sets such as Multi-PIE. We demonstrate how to capture a set of training images with enough illumination variation that they span test images taken under uncontrolled illumination. In order to evaluate how our algorithms work under practical testing conditions, we have implemented a complete face recognition system, including a projector-based training acquisition system. Our system can efficiently and effectively recognize faces under a variety of realistic conditions, using only frontal images under the proposed illuminations as training.

Original languageEnglish (US)
Article number5871642
Pages (from-to)372-386
Number of pages15
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume34
Issue number2
DOIs
StatePublished - Jan 1 2012

Fingerprint

Sparse Representation
Face recognition
Face Recognition
Illumination
Alignment
Lighting
Misalignment
Face
Occlusion
Testing
Recognition Algorithm
Projector
Training
Robustness
Partial
Scenarios

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Vision and Pattern Recognition
  • Computational Theory and Mathematics
  • Artificial Intelligence
  • Applied Mathematics

Cite this

Wagner, Andrew ; Wright, John ; Ganesh, Arvind ; Zhou, Zihan ; Mobahi, Hossein ; Ma, Yi. / Toward a practical face recognition system : Robust alignment and illumination by sparse representation. In: IEEE Transactions on Pattern Analysis and Machine Intelligence. 2012 ; Vol. 34, No. 2. pp. 372-386.
@article{8c4a571223d64536ba1dae66f9944caa,
title = "Toward a practical face recognition system: Robust alignment and illumination by sparse representation",
abstract = "Many classic and contemporary face recognition algorithms work well on public data sets, but degrade sharply when they are used in a real recognition system. This is mostly due to the difficulty of simultaneously handling variations in illumination, image misalignment, and occlusion in the test image. We consider a scenario where the training images are well controlled and test images are only loosely controlled. We propose a conceptually simple face recognition system that achieves a high degree of robustness and stability to illumination variation, image misalignment, and partial occlusion. The system uses tools from sparse representation to align a test face image to a set of frontal training images. The region of attraction of our alignment algorithm is computed empirically for public face data sets such as Multi-PIE. We demonstrate how to capture a set of training images with enough illumination variation that they span test images taken under uncontrolled illumination. In order to evaluate how our algorithms work under practical testing conditions, we have implemented a complete face recognition system, including a projector-based training acquisition system. Our system can efficiently and effectively recognize faces under a variety of realistic conditions, using only frontal images under the proposed illuminations as training.",
author = "Andrew Wagner and John Wright and Arvind Ganesh and Zihan Zhou and Hossein Mobahi and Yi Ma",
year = "2012",
month = "1",
day = "1",
doi = "10.1109/TPAMI.2011.112",
language = "English (US)",
volume = "34",
pages = "372--386",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
issn = "0162-8828",
publisher = "IEEE Computer Society",
number = "2",

}

Toward a practical face recognition system : Robust alignment and illumination by sparse representation. / Wagner, Andrew; Wright, John; Ganesh, Arvind; Zhou, Zihan; Mobahi, Hossein; Ma, Yi.

In: IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 34, No. 2, 5871642, 01.01.2012, p. 372-386.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Toward a practical face recognition system

T2 - Robust alignment and illumination by sparse representation

AU - Wagner, Andrew

AU - Wright, John

AU - Ganesh, Arvind

AU - Zhou, Zihan

AU - Mobahi, Hossein

AU - Ma, Yi

PY - 2012/1/1

Y1 - 2012/1/1

N2 - Many classic and contemporary face recognition algorithms work well on public data sets, but degrade sharply when they are used in a real recognition system. This is mostly due to the difficulty of simultaneously handling variations in illumination, image misalignment, and occlusion in the test image. We consider a scenario where the training images are well controlled and test images are only loosely controlled. We propose a conceptually simple face recognition system that achieves a high degree of robustness and stability to illumination variation, image misalignment, and partial occlusion. The system uses tools from sparse representation to align a test face image to a set of frontal training images. The region of attraction of our alignment algorithm is computed empirically for public face data sets such as Multi-PIE. We demonstrate how to capture a set of training images with enough illumination variation that they span test images taken under uncontrolled illumination. In order to evaluate how our algorithms work under practical testing conditions, we have implemented a complete face recognition system, including a projector-based training acquisition system. Our system can efficiently and effectively recognize faces under a variety of realistic conditions, using only frontal images under the proposed illuminations as training.

AB - Many classic and contemporary face recognition algorithms work well on public data sets, but degrade sharply when they are used in a real recognition system. This is mostly due to the difficulty of simultaneously handling variations in illumination, image misalignment, and occlusion in the test image. We consider a scenario where the training images are well controlled and test images are only loosely controlled. We propose a conceptually simple face recognition system that achieves a high degree of robustness and stability to illumination variation, image misalignment, and partial occlusion. The system uses tools from sparse representation to align a test face image to a set of frontal training images. The region of attraction of our alignment algorithm is computed empirically for public face data sets such as Multi-PIE. We demonstrate how to capture a set of training images with enough illumination variation that they span test images taken under uncontrolled illumination. In order to evaluate how our algorithms work under practical testing conditions, we have implemented a complete face recognition system, including a projector-based training acquisition system. Our system can efficiently and effectively recognize faces under a variety of realistic conditions, using only frontal images under the proposed illuminations as training.

UR - http://www.scopus.com/inward/record.url?scp=84055212058&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84055212058&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2011.112

DO - 10.1109/TPAMI.2011.112

M3 - Article

C2 - 21646680

AN - SCOPUS:84055212058

VL - 34

SP - 372

EP - 386

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

IS - 2

M1 - 5871642

ER -