Sum of Sine Modeling Approach as a New Processing Technique For a Biometric System Based on ECG Signal

Abstract

Biometrics is considered in current research as one of the best methods for authenticating human beings. In our paper, the heartbeat biometric, also called Electrocardiographic (ECG), is working on. This biometric is chosen because human ECGs cannot be falsely created and replicated. This study aims to find the best features from this biometric that can identify a person, given the extractions and classification algorithms for the heartbeat biometric signal. Depending on a literature study we work to propose a new and more efficient technique based on a new method for ECG features extraction and these features will be the inputs for pattern recognition classifier. This methodology will be tested on real experimental ECG data that is collected. The Data collected from 10 subjects by a commercial ECG device taking the data from lead 1. The pre-processing steps start with the Empirical Mode Decomposition (EMD) before digital filters which are: low pass, high pass, and derivative pass filters. Features extraction steps are peak detection, segmentation, and wave modeling for each segment. The classification used the Multi-Layer Perceptron and compared it to classification using Radial Basis Function were the results of MLP were much better for these applications since the accuracy of the final results of MLP is 99% and that related to the RBF is 95%.