How effective are landmarks and their geometry for face recognition?

J. Shi, Ashok K Samal, D. Marx

Research output: Contribution to journalArticle

97 Citations (Scopus)

Abstract

This paper evaluates how biologically meaningful landmarks and their geometry extracted from face images can be used for face recognition. The traditional Procrustes distance is studied for the landmark-based face model. By using complex principal component analysis, we propose a refined Procrustes distance that incorporates statistical correlation of landmarks. Motivated by research results from art, anthropology, and aesthetic surgery where ratios play a significant role in human face descriptions, we also investigate how well-normalized Euclidean distances (special ratios) can be exploited for face recognition. Exploiting symmetry and using principal component analysis, we reduce the number of ratios to 20. In this investigation, we analyze the effectiveness of three well-known similarity measures including the typical l1 norm, l2 norm, and Mahalanobis distance. We also define a new similarity measure called the eigenvalue-weighted cosine (EWC) distance. We evaluate our approach using two standard face databases: the Purdue AR and the FERET database. The former has well-controlled face images and we use it to test the reliability of landmarks. The FERET database is used for performance evaluation of face recognition algorithms. This papers use a performance measure called the cumulative match score, which indicates the percentage of faces that can be eliminated from consideration. Experimental results show that our EWC distance outperforms the l1 and l2 norms and even the Mahalanobis distance. Our study also finds that landmarks and their geometry-based approach can account for variations of face expression and aging very well. Thus, they can be used either in stand-alone mode or in conjunction with other approaches to reduce the search space a priori.

Original languageEnglish (US)
Pages (from-to)117-133
Number of pages17
JournalComputer Vision and Image Understanding
Volume102
Issue number2
DOIs
StatePublished - May 1 2006

Fingerprint

Face recognition
Principal component analysis
Geometry
Surgery
Aging of materials

Keywords

  • FERET
  • Face recognition
  • Landmarks
  • Procrustes distance
  • Similarity measure

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition

Cite this

How effective are landmarks and their geometry for face recognition? / Shi, J.; Samal, Ashok K; Marx, D.

In: Computer Vision and Image Understanding, Vol. 102, No. 2, 01.05.2006, p. 117-133.

Research output: Contribution to journalArticle

@article{ebc834e7337d4d669ffed5378d10e953,
title = "How effective are landmarks and their geometry for face recognition?",
abstract = "This paper evaluates how biologically meaningful landmarks and their geometry extracted from face images can be used for face recognition. The traditional Procrustes distance is studied for the landmark-based face model. By using complex principal component analysis, we propose a refined Procrustes distance that incorporates statistical correlation of landmarks. Motivated by research results from art, anthropology, and aesthetic surgery where ratios play a significant role in human face descriptions, we also investigate how well-normalized Euclidean distances (special ratios) can be exploited for face recognition. Exploiting symmetry and using principal component analysis, we reduce the number of ratios to 20. In this investigation, we analyze the effectiveness of three well-known similarity measures including the typical l1 norm, l2 norm, and Mahalanobis distance. We also define a new similarity measure called the eigenvalue-weighted cosine (EWC) distance. We evaluate our approach using two standard face databases: the Purdue AR and the FERET database. The former has well-controlled face images and we use it to test the reliability of landmarks. The FERET database is used for performance evaluation of face recognition algorithms. This papers use a performance measure called the cumulative match score, which indicates the percentage of faces that can be eliminated from consideration. Experimental results show that our EWC distance outperforms the l1 and l2 norms and even the Mahalanobis distance. Our study also finds that landmarks and their geometry-based approach can account for variations of face expression and aging very well. Thus, they can be used either in stand-alone mode or in conjunction with other approaches to reduce the search space a priori.",
keywords = "FERET, Face recognition, Landmarks, Procrustes distance, Similarity measure",
author = "J. Shi and Samal, {Ashok K} and D. Marx",
year = "2006",
month = "5",
day = "1",
doi = "10.1016/j.cviu.2005.10.002",
language = "English (US)",
volume = "102",
pages = "117--133",
journal = "Computer Vision and Image Understanding",
issn = "1077-3142",
publisher = "Academic Press Inc.",
number = "2",

}

TY - JOUR

T1 - How effective are landmarks and their geometry for face recognition?

AU - Shi, J.

AU - Samal, Ashok K

AU - Marx, D.

PY - 2006/5/1

Y1 - 2006/5/1

N2 - This paper evaluates how biologically meaningful landmarks and their geometry extracted from face images can be used for face recognition. The traditional Procrustes distance is studied for the landmark-based face model. By using complex principal component analysis, we propose a refined Procrustes distance that incorporates statistical correlation of landmarks. Motivated by research results from art, anthropology, and aesthetic surgery where ratios play a significant role in human face descriptions, we also investigate how well-normalized Euclidean distances (special ratios) can be exploited for face recognition. Exploiting symmetry and using principal component analysis, we reduce the number of ratios to 20. In this investigation, we analyze the effectiveness of three well-known similarity measures including the typical l1 norm, l2 norm, and Mahalanobis distance. We also define a new similarity measure called the eigenvalue-weighted cosine (EWC) distance. We evaluate our approach using two standard face databases: the Purdue AR and the FERET database. The former has well-controlled face images and we use it to test the reliability of landmarks. The FERET database is used for performance evaluation of face recognition algorithms. This papers use a performance measure called the cumulative match score, which indicates the percentage of faces that can be eliminated from consideration. Experimental results show that our EWC distance outperforms the l1 and l2 norms and even the Mahalanobis distance. Our study also finds that landmarks and their geometry-based approach can account for variations of face expression and aging very well. Thus, they can be used either in stand-alone mode or in conjunction with other approaches to reduce the search space a priori.

AB - This paper evaluates how biologically meaningful landmarks and their geometry extracted from face images can be used for face recognition. The traditional Procrustes distance is studied for the landmark-based face model. By using complex principal component analysis, we propose a refined Procrustes distance that incorporates statistical correlation of landmarks. Motivated by research results from art, anthropology, and aesthetic surgery where ratios play a significant role in human face descriptions, we also investigate how well-normalized Euclidean distances (special ratios) can be exploited for face recognition. Exploiting symmetry and using principal component analysis, we reduce the number of ratios to 20. In this investigation, we analyze the effectiveness of three well-known similarity measures including the typical l1 norm, l2 norm, and Mahalanobis distance. We also define a new similarity measure called the eigenvalue-weighted cosine (EWC) distance. We evaluate our approach using two standard face databases: the Purdue AR and the FERET database. The former has well-controlled face images and we use it to test the reliability of landmarks. The FERET database is used for performance evaluation of face recognition algorithms. This papers use a performance measure called the cumulative match score, which indicates the percentage of faces that can be eliminated from consideration. Experimental results show that our EWC distance outperforms the l1 and l2 norms and even the Mahalanobis distance. Our study also finds that landmarks and their geometry-based approach can account for variations of face expression and aging very well. Thus, they can be used either in stand-alone mode or in conjunction with other approaches to reduce the search space a priori.

KW - FERET

KW - Face recognition

KW - Landmarks

KW - Procrustes distance

KW - Similarity measure

UR - http://www.scopus.com/inward/record.url?scp=33645821651&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33645821651&partnerID=8YFLogxK

U2 - 10.1016/j.cviu.2005.10.002

DO - 10.1016/j.cviu.2005.10.002

M3 - Article

AN - SCOPUS:33645821651

VL - 102

SP - 117

EP - 133

JO - Computer Vision and Image Understanding

JF - Computer Vision and Image Understanding

SN - 1077-3142

IS - 2

ER -