Multimodal Biometric Identification System Based EEG (Electroencephalograph) and Fingerprint with Template Protection

Abstract

This research presents a multimodal biometric system that integrates EEG and fingerprint data using deep learning convolutional neural networks. The system addresses the limitations of unimodal biometric systems and enhances template security by implementing a fuzzy vault scheme. The system extracts frequency weighted power (WFP) features from EEG data and minutiae from fingerprints, combines them, and uses the combined features along with a secret key to create a database in the vault. Experimental results demonstrate that the proposed system outperforms other methods, achieving an Equal Error Rate (EER) of 0.25% for EEG, 0.20% for fingerprint, and 0.10% for multimodal unlike Liwen, (2010) with an Equal Error Rate (EER of 1.12%). The fuzzy vault biometric system also performed exceptionally well, achieving perfect accuracy (EER of 0.00) in differentiating between genuine and impostor samples, with a perfect ROC AUC value of 1.00 Unlike Suputra and Sukarno, (2019) with False Rejection Rate (FRR) of 8.9475% and False Rejection Rate (FAR) of 0.3520% equivalent to an Equal Error Rate (EER Of 0.045). The t-test analysis confirms that the difference in scores is statistically significant, providing further evidence of the system's robust performance. Overall, these results suggest that the fuzzy vault implementation is performing exceptionally well in terms of security and accuracy. This study is significant because it proposes a new method for fusion normalization of EEG brain signal and fingerprint with template protection scheme using fuzzy vault to provide better accuracy and high template protection. The feature work should use large data set for both the EEG and Fingerprint and also should use another model for classifier based score normalization.