Design of Heart Sound Biometric Access Control System

Abstract

This project reviews the extent to using of heart sound as biometric system used to identify the person and verify him. Features that have been chosen for this purpose are energy of heart sound signal and power spectrum.The heart sounds were collected from the internet. A matrix laboratory (MATLAB) algorithm was implemented to design a program that used discrete wavelet transform (DWT) for de-noising the signal, continuous wavelet transform (CWT) and Mel Frequency Cepstrum Coefficients (MFCC) for feature extraction and then a databases were build holding sounds features. Finally, Euclidean distance was used for classification sounds and display result depending on mode of operation (identification or verification mode). The program code was implemented in Graphical User Interface (GUI) to easy use of application. Then RS232 was interfaced between it and simulation circuit that consist of ATMEGA 16 (microcontroller), which simulate program via Bascom-AVR as programming language then Proteus was used to verify the code, it was displayed the name of person if the application in identification mode or allowed to access if in verification mode. The application which was programmed and simulated gave a 22.35% error in identification mode when used energy signature and gave a 15.29% error in identification mode when used power spectrum signature.