Statistical transformations of frontal models for non-frontal face verification

Sanderson, C. and Bengio, S. (2004). Statistical transformations of frontal models for non-frontal face verification. In: Image Processing 2004 (ICIP 2004). International Conference on Image Processing (ICIP 2004), Singapore, (585-588). 24-27 October 2004. doi:10.1109/ICIP.2004.1418822


Author Sanderson, C.
Bengio, S.
Title of paper Statistical transformations of frontal models for non-frontal face verification
Conference name International Conference on Image Processing (ICIP 2004)
Conference location Singapore
Conference dates 24-27 October 2004
Proceedings title Image Processing 2004 (ICIP 2004)   Check publisher's open access policy
Journal name Icip: 2004 International Conference On Image Processing, Vols 1- 5   Check publisher's open access policy
Place of Publication Piscataway, NJ, U.S.A.
Publisher IEEE - Institute of Electrical Electronics Engineers Inc.
Publication Year 2004
Sub-type Fully published paper
DOI 10.1109/ICIP.2004.1418822
Open Access Status Not yet assessed
ISBN 0-7803-8554-3
ISSN 1522-4880
Volume 1
Start page 585
End page 588
Total pages 4
Language eng
Abstract/Summary In the framework of a face Verification System using local feature and a Gaussian Mixture Model based classifier, we address the problem of non-frontal face verification (when only a single (frontal) training, image is available) by extending each client's frontal face model with artificially synthesized models for non-frontal views. Furthermore. we propose the Maximum Likelihood Shift (MLS) synthesis technique and compare its performance against a Maximum Likelihood Linear Regression (MLLR) based technique (orginally developed for adapting speech recognition systems) and the recently proposed "difference between two Universal Background Models" (UBMdiff) technique. All techniques rely on prior information and learn how a generic face model for the frontal view is related to generic models at non-frontal views. Experiments on the FERET database suggest that that the proposed MLS technique is more Suitable than MLLR (due to a lower number of free parameters) and UBMdiff (due to lack of heuristics). The results further suggest that extending frontal models considerably reduces errors.
Subjects 080104 Computer Vision
080106 Image Processing
080109 Pattern Recognition and Data Mining
010401 Applied Statistics
Keyword Face verification system
Gaussian mixture model based classifier
Maximum likelihood shift (MLS)
Maximum likelihood linear regression (MLLR)
Q-Index Code E1

 
Versions
Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 0 times in Thomson Reuters Web of Science Article
Scopus Citation Count Cited 3 times in Scopus Article | Citations
Google Scholar Search Google Scholar
Created: Wed, 01 Apr 2009, 20:01:00 EST by Laura McTaggart on behalf of School of Information Technol and Elec Engineering