Face recognition by statistical analysis of feature detectors

Citation
P. Kalocsai et al., Face recognition by statistical analysis of feature detectors, IMAGE VIS C, 18(4), 2000, pp. 273-278
Citations number
8
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
IMAGE AND VISION COMPUTING
ISSN journal
02628856 → ACNP
Volume
18
Issue
4
Year of publication
2000
Pages
273 - 278
Database
ISI
SICI code
0262-8856(20000301)18:4<273:FRBSAO>2.0.ZU;2-#
Abstract
A successful face recognition system calculates similarity of face images b ased on the activation of multiscale and multiorientation Gabor kernels, bu t without utilizing any statistical properties of the given face data [M. L ades, J.C. Vortbruggen, J. Buhmann, J. Lange, C. von der Malsburg, R.P. Wur tz, W. Konen, Distortion invariant object recognition in the dynamic link a rchitecture, IEEE Transactions on Computers 42 (1993) 300-311]. A method ha s been developed to weight the contribution of each element (1920 kernels) in the representation according to its power of predicting similarity of fa ces. The same statistical method has also been used to assess how changes i n orientation (horizontal and vertical), expression, illumination and backg round contribute to the overall variance in the kernel activations. It was shown on a Caucasian and a Japanese image-set that weighting the elements i n the representation according to their discriminative power would increase recognition performance;It has also been demonstrated that the weighting m ethod is particularly useful when data compression is a key requirement. Th e advantages of the weighting scheme were also verified by double cross-val idation. (C) 2000 Elsevier Science B.V. All rights reserved.