Intelligent Multi-Modal Biometric Authentication with Face, Iris, and Palmprint Fusion Using ANFIS

Abstract

In the evolving landscape of cybersecurity, traditional single-modal biometric authentication systems face significant challenges such as spoofing attacks and performance degradation due to environmental factors. To address these vulnerabilities, this paper presents an Intelligent Multi-Modal Biometric Authentication System that combines face, iris, and palmprint biometrics for robust human verification. The proposed system employs an Adaptive Neuro-Fuzzy Inference System (ANFIS) for intelligent fusion of biometric features and matching scores, enhancing decision-making and classification accuracy in diverse authentication scenarios.

Biometric data is captured live using a standard webcam under white-light conditions for all three modalities. Feature extraction is performed using domain-specific techniques, and individual matching scores are normalized. To combat spoofing, liveness detection modules are integrated—blink detection for the face and pupil dilation tracking for the iris—ensuring the subject is alive and present during authentication.

Three fusion strategies are implemented and evaluated: feature-level fusion, score-level fusion, and hybrid fusion. Experimental results demonstrate that the hybrid fusion method achieves superior performance, with an accuracy of 98.1%, a false acceptance rate (FAR) of 0.9%, and a false rejection rate (FRR) of 1.5%.

To the best of our knowledge, this is the first implementation that integrates ANFIS-based hybrid fusion with on-device liveness detection across face, iris, and palmprint modalities using a single low-cost webcam.

The system’s performance is validated using both public datasets and a real-world dataset collected from 20 users. The results confirm that the proposed framework offers a secure, scalable, and efficient solution for high-assurance biometric authentication in organizational environments.