Analyzing The Effect of Segmentation on PCG Based Biometric System

A new comparative approach to analyzing the effect of segmentation on PCG-based biometrics is proposed here. In this HSCT-11, a dataset is used which contains data that was collected from 206 people. First, PCG signals were pre-processed to remove noise and other artifacts using Discrete wavelet transform (DWT) up to the Fifth level, and after reconstruction of the signal nine-time and frequency domain features were selected to do various experiments using three different types of classifiers and found out that DNN gives a good response. A data set of 40 different person samples that are split into 11,468 samples is used here and an accuracy of 85.87% is achieved using DNN.