Presentation 2001/11/8
Person recognition based on the integration of speech and facial image with Dempster-Shafer's theory
Yoshiaki SUGIE, Tetsunori KOBAYASHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we describe a new integration method of speech and facial image information for person recognition. We show that the integration of speech and facial image with Dempster-Shafer's theory is very effective to improve the total performance of media-integrated recognizer especially to improve the rejection performance. Conventional media-integrated recognizers adopt the simple sum of the normalized log-likelihood values of uni-modal Bayesian recognizers and use it alone for the decision. We modified the method based on Dempster-Shafer's theory to solve the problem of Bayesian probability that can't use uncertainty. Dempster-Shafer's theory gives the attractive ideas for the integration of numerical evidences, but there was no concrete and reasonable idea for enumerating the each evidence. We solved this problem by considering the cumulative density functions of normalized log-likelihood values of correct and incorrect categories. The experimental results shows that proposed method improves the recognition performance by 18.0% compared to conventional method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Person recognition / Embedded HMM-based face recognition / GMM-based speaker recognition / Information integration / Dempster-Shafer theory
Paper # PRMU2001-122,HIP2001-25,MVE2001-84
Date of Issue

Conference Information
Committee MVE
Conference Date 2001/11/8(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Media Experience and Virtual Environment (MVE)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Person recognition based on the integration of speech and facial image with Dempster-Shafer's theory
Sub Title (in English)
Keyword(1) Person recognition
Keyword(2) Embedded HMM-based face recognition
Keyword(3) GMM-based speaker recognition
Keyword(4) Information integration
Keyword(5) Dempster-Shafer theory
1st Author's Name Yoshiaki SUGIE
1st Author's Affiliation Department of Electrical, Electronics & Computer Engineering Waseda University()
2nd Author's Name Tetsunori KOBAYASHI
2nd Author's Affiliation Department of Electrical, Electronics & Computer Engineering Waseda University
Date 2001/11/8
Paper # PRMU2001-122,HIP2001-25,MVE2001-84
Volume (vol) vol.101
Number (no) 425
Page pp.pp.-
#Pages 6
Date of Issue