Neighbourhood discriminant locally linear embedding in face recognition

Pang Ying Han, Andrew Teoh Beng Jin, Wong Eng Kiong

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

9 Citations (Scopus)

Abstract

Face images are often very high-dimensional and complex. However, the actual underlying structure can be characterized by a small number of features. Hence, locally linear embedding (LIE) is proposed as a nonlinear dimension reduction technique to deal this problem. LLE learns the intrinsic manifold embedded in the high dimensional ambient space by minimizing the global reconstruction error of the neighbourhood in the data set. LLE is popular in analyzing face images with different poses, illuminations or facial expressions for one subject class. It is developed based on the assumption that data that is distributed on a single manifold is having the same class label; hence the process of neighborhood selection is non class-specific. However, this is inappropriate to face recognition as face recognition learns in multiple manifolds where each representing data on one specific class. Here, we modify the original LLE by embedding prior class information in the process of neighborhood selection. Experimental results demonstrate that our technique consistently outperforms the original LLE in ORL, PLE and FRGC databases.

Original languageEnglish
Title of host publicationProceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV
PublisherIEEE Computer Society
Pages223-228
Number of pages6
ISBN (Print)0769533590, 9780769533599
DOIs
Publication statusPublished - 2008 Jan 1
Event5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV - Penang, Malaysia
Duration: 2008 Aug 262008 Aug 28

Publication series

NameProceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV

Other

Other5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV
CountryMalaysia
CityPenang
Period08/8/2608/8/28

Fingerprint

Face recognition
Labels
Lighting

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications
  • Computer Science Applications
  • Computer Vision and Pattern Recognition

Cite this

Han, P. Y., Beng Jin, A. T., & Kiong, W. E. (2008). Neighbourhood discriminant locally linear embedding in face recognition. In Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV (pp. 223-228). [4627011] (Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV). IEEE Computer Society. https://doi.org/10.1109/CGIV.2008.63
Han, Pang Ying ; Beng Jin, Andrew Teoh ; Kiong, Wong Eng. / Neighbourhood discriminant locally linear embedding in face recognition. Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV. IEEE Computer Society, 2008. pp. 223-228 (Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV).
@inproceedings{60d2d5a735a34b7fb2df9ffbf2ef5c03,
title = "Neighbourhood discriminant locally linear embedding in face recognition",
abstract = "Face images are often very high-dimensional and complex. However, the actual underlying structure can be characterized by a small number of features. Hence, locally linear embedding (LIE) is proposed as a nonlinear dimension reduction technique to deal this problem. LLE learns the intrinsic manifold embedded in the high dimensional ambient space by minimizing the global reconstruction error of the neighbourhood in the data set. LLE is popular in analyzing face images with different poses, illuminations or facial expressions for one subject class. It is developed based on the assumption that data that is distributed on a single manifold is having the same class label; hence the process of neighborhood selection is non class-specific. However, this is inappropriate to face recognition as face recognition learns in multiple manifolds where each representing data on one specific class. Here, we modify the original LLE by embedding prior class information in the process of neighborhood selection. Experimental results demonstrate that our technique consistently outperforms the original LLE in ORL, PLE and FRGC databases.",
author = "Han, {Pang Ying} and {Beng Jin}, {Andrew Teoh} and Kiong, {Wong Eng}",
year = "2008",
month = "1",
day = "1",
doi = "10.1109/CGIV.2008.63",
language = "English",
isbn = "0769533590",
series = "Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV",
publisher = "IEEE Computer Society",
pages = "223--228",
booktitle = "Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV",
address = "United States",

}

Han, PY, Beng Jin, AT & Kiong, WE 2008, Neighbourhood discriminant locally linear embedding in face recognition. in Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV., 4627011, Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV, IEEE Computer Society, pp. 223-228, 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV, Penang, Malaysia, 08/8/26. https://doi.org/10.1109/CGIV.2008.63

Neighbourhood discriminant locally linear embedding in face recognition. / Han, Pang Ying; Beng Jin, Andrew Teoh; Kiong, Wong Eng.

Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV. IEEE Computer Society, 2008. p. 223-228 4627011 (Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV).

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

TY - GEN

T1 - Neighbourhood discriminant locally linear embedding in face recognition

AU - Han, Pang Ying

AU - Beng Jin, Andrew Teoh

AU - Kiong, Wong Eng

PY - 2008/1/1

Y1 - 2008/1/1

N2 - Face images are often very high-dimensional and complex. However, the actual underlying structure can be characterized by a small number of features. Hence, locally linear embedding (LIE) is proposed as a nonlinear dimension reduction technique to deal this problem. LLE learns the intrinsic manifold embedded in the high dimensional ambient space by minimizing the global reconstruction error of the neighbourhood in the data set. LLE is popular in analyzing face images with different poses, illuminations or facial expressions for one subject class. It is developed based on the assumption that data that is distributed on a single manifold is having the same class label; hence the process of neighborhood selection is non class-specific. However, this is inappropriate to face recognition as face recognition learns in multiple manifolds where each representing data on one specific class. Here, we modify the original LLE by embedding prior class information in the process of neighborhood selection. Experimental results demonstrate that our technique consistently outperforms the original LLE in ORL, PLE and FRGC databases.

AB - Face images are often very high-dimensional and complex. However, the actual underlying structure can be characterized by a small number of features. Hence, locally linear embedding (LIE) is proposed as a nonlinear dimension reduction technique to deal this problem. LLE learns the intrinsic manifold embedded in the high dimensional ambient space by minimizing the global reconstruction error of the neighbourhood in the data set. LLE is popular in analyzing face images with different poses, illuminations or facial expressions for one subject class. It is developed based on the assumption that data that is distributed on a single manifold is having the same class label; hence the process of neighborhood selection is non class-specific. However, this is inappropriate to face recognition as face recognition learns in multiple manifolds where each representing data on one specific class. Here, we modify the original LLE by embedding prior class information in the process of neighborhood selection. Experimental results demonstrate that our technique consistently outperforms the original LLE in ORL, PLE and FRGC databases.

UR - http://www.scopus.com/inward/record.url?scp=55349133424&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=55349133424&partnerID=8YFLogxK

U2 - 10.1109/CGIV.2008.63

DO - 10.1109/CGIV.2008.63

M3 - Conference contribution

AN - SCOPUS:55349133424

SN - 0769533590

SN - 9780769533599

T3 - Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV

SP - 223

EP - 228

BT - Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV

PB - IEEE Computer Society

ER -

Han PY, Beng Jin AT, Kiong WE. Neighbourhood discriminant locally linear embedding in face recognition. In Proceedings - 5th International Conference on Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV. IEEE Computer Society. 2008. p. 223-228. 4627011. (Proceedings - Computer Graphics, Imaging and Visualisation, Modern Techniques and Applications, CGIV). https://doi.org/10.1109/CGIV.2008.63