An ensemble approach for electrocardiogram and lip features based biometric authentication by using grey wolf optimization

Abstract

In the pursuit of fortified security measures, the convergence of multimodal biometric authentication and ensemble learning techniques have emerged as a pivotal domain of research. This study explores the integration of multimodal biometric authentication and ensemble learning techniques to enhance security. Focusing on lip movement and electrocardiogram (ECG) data, the research combines their distinct characteristics for advanced authentication. Ensemble learning merges diverse models, achieving increased accuracy and resilience in multimodal fusion. Harmonizing lip and ECG modalities establishes a robust authentication system, countering vulnerabilities in unimodal methods. This approach leverages ECG's robustness against spoofing attacks and lip's fine-grained behavioral cues for comprehensive authentication. Ensemble learning techniques, from majority voting to advanced methods, harness the strengths of individual models, improving accuracy, reliability, and generalization. Moreover, ensemble learning detects anomalies, enhancing security. The study incorporates ECG signal filtering and lip region extraction as preprocessing, uses wavelet transform for ECG features, SIFT for lip image features, and employs grey-wolf optimization for feature selection. Ultimately, a voting-based ensemble classifier is applied for classification, showcasing the potential of this integrated approach in fortified security measures.