Person Identification based on ECG Signals using Continuous Wavelet Transform

The development of biometric methods has been given significant importance in recent years. Accurate and unassuming identification and authentication of persons is a crucial fundament in protecting personal or industrial property. This work presents the person identification based on electrocardiographic (ECG) signals, where several approaches are investigated. The own database of ECG signals acquired from 19 subjects is created for this purpose. Signals are split utilizing blind segmentation and R peak centered segmentation with four different segment lengths. They are further represented in the time-frequency domain in the form of scalograms, incorporating the Morse, Morlet, and Bump wavelet. The convolutional neural network (CNN) models are proposed for the person identification, while the highest average classification accuracy of 99.60% is achieved.