A Novel Multimodal Biometric Person Authentication System Based on ECG and Iris Data.

Abstract

Existing security issues like keys, pins, and passwords employed presently in almost all the fields that have certain limitations like passwords and pins can be easily forgotten; keys can be lost. To overcome such security issues, new biometric features have shown outstanding improvements in authentication systems as a result of significant developments in biological digital signal processing. Currently, the multimodal authentications have gained huge attention in biometric systems which can be either behavioural or physiological. A biometric system with multimodality club data from many biometric modalities increases each biometric system's performance and makes it more resistant to spoof attempts. Apart from electrocardiogram (ECG) and iris, there are a lot of other biometric traits that can be captured from the human body. They include face, fingerprint, gait, keystroke dynamics, voice, DNA, palm vein, and hand geometry recognition. Electrocardiograms (ECG) have recently been employed in unimodal and multimodal biometric recognition systems as a novel biometric technology. When compared to other biometric approaches, ECG has the intrinsic quality of a person's liveness, making it difficult to fake. Similarly, the iris also plays an important role in biometric authentication. Based on these assumptions, we present a multimodal biometric person authentication system. The projected method includes preprocessing, segmentation, feature extraction, feature fusion, and ensemble classifier where majority voting is presented to obtain the final outcome. The comparative analysis shows the overall performance as 96.55%, 96.2%, 96.2%, 96.5%, and 95.65% in terms of precision, F1-score, sensitivity, specificity, and accuracy.