Biometric Recognition for Twins Inconsideration of Age Variability Using PPG Signals

Abstract

Recent biometric modalities involve biomedical signals such as PPG to identify individuals. This study has been motivated by this new research area using PPG signal to identify twins incorporating age variability. The proposed system is suggested to be a substitute to the current traditional methods being used widely nowadays. A total of 21 subjects were used for experimentation purposes and lowpass filter is applied to remove unwanted noise from the signal. Distinctive features are extracted from the filtered PPG signals. Later, Bayes Network (BN), Naïve Bayes (NB), Radial Basis Function (RBF) and Multilayer Perceptron (MLP) were used to classify the subjects using the discriminant features. Based on the experimentation results, classification accuracies ranging from 90% to 100% were achieved by categorizing the data into six different sets which are overall dataset, Groups I, II, III, IV and V. The result provides an alternative mechanism to identify twins using PPG signals incorporating age variability besides using the traditional methods.