74 International Journal of Systems Biology and Biomedical Technologies, 1(4), 74-87, October-December 2012
Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Keywords: Feature Extraction, Gaussian Mixture Model, Heart Sounds, Human Print, Identifcation,
Vector Quantization, Verifcation
1. INTRODUCTION
The need to identify persons correctly and
irrevocably has existed for a very long time.
The authorization to enter a building, to open a
cupboard, to cross a border, to get money from a
bank etc. is always connected to the identity of
a person. It is therefore necessary to prove this
identity in one way or another. This procedure
is called verification. A person claims to be
authorized or to have a certain identity, and this
must then be verified (Wahid et al., in press).
Heart Sounds Human
Identifcation and Verifcation
Approaches using Vector
Quantization and Gaussian
Mixture Models
Neveen I. Ghali, Faculty of Science, Al-Azhar University, Cairo, Egypt
Rasha Wahid, Faculty of Science, Al-Azhar University, Cairo, Egypt
Aboul Ella Hassanien, Scientifc Research Group in Egypt (SRGE), Faculty of Computers and
Information, Cairo University, Cairo, Egypt
ABSTRACT
In this paper the possibility of using the human heart sounds as a human print is investigated. To evaluate the
performance and the uniqueness of the proposed approach, tests using a high resolution auscultation digital
stethoscope are done for nearly 80 heart sound samples. The verifcation approach consists of a robust feature
extraction with a specifed confguration in conjunction with Gaussian mixture modeling. The similarity of
two samples is estimated by measuring the difference between their negative log-likelihood similarities of the
features. The experimental results obtained show that the overall accuracy offered by the employed Gaussian
mixture modeling reach up to 85%. The identifcation approach consists of a robust feature extraction with a
specifed confguration in conjunction with LBG-VQ. The experimental results obtained show that the overall
accuracy offered by the employed LBG-VQ reach up to 88.7%.
DOI: 10.4018/ijsbbt.2012100106