Abstract:
This thesis studies the applicability of the electrocardiogram signal (ECG) as a biometric.
The ECG, a record of electrical currents generated by the beating heart, is potentially a
distinctive human characteristic, since ECG waveforms and other properties of the ECG
depend on the anatomic features of the human heart and body. This thesis, propose an approach
based on a combination of time domain ’fiducial’ based features (R-peak amplitude and
power) and a wavelet ’non–fiducial’ based features (energy and power) for automatic analysis
of the ECG for application in human recognition. A set of 26 ECG data were obtained from
four publicly available databases, namely the MIT-BIH Normal Sinus Rhythm, ECG-ID
Database, MITd BIH Long Term Database and the MIT-BIH ST Change were used to
evaluate the proposed system. The methodology consists of three major stages: preprocessing,
feature extraction and classification. For noise reduction, the input ECG signal
was first denoised using Discrete Wavelet Transform (DWT). The four features extracted from
each subject. Finally, the recognition concluded by classifying the extracted features using
Linear Discriminant Analysis (LDA).The proposed algorithm achieved accuracy of 81.3%
based on time domain features and accuracy of 83.3% based on wavelet features .The
recognition accuracy achieved 100% when both wavelet and time domain features were
combined.
