Fusion-Enhanced Hybrid Multimodal Biometric System: Integrating Visible and Infrared Facial Recognition for Robust Authentication
Mohamed Abdul‐Al, George Kumi Kyeremeh, Rami Qahwaji, Nazar T. Ali, Raed A. Abd‐Alhameed · IEEE Access · 2026
Biometric authentication is a pivotal technology for secure identification and access control, harnessing unique physical or behavioural traits. Traditional unimodal biometric systems, while widely implemented, face significant challenges, including susceptibility to noise, non-universality, and spoofing. This study presents an innovative hybrid multimodal biometric system (MBS) that integrates visible (VIS) and infrared (IR) facial imagery, with experimental implementation restricted to weighted score-level fusion, to address these limitations. Leveraging VGG16 for feature extraction, Principal Component Analysis (PCA) for dimensionality reduction and Sequential Neural Networks (NNs) for Classification, the system employs advanced weighted score-level fusion techniques to enhance robustness and accuracy for both verification (matching a claimed identity) and identification (matching against all stored identities). Evaluated on the Sejong Face Dataset (SFD) under diverse conditions, the proposed approach achieves an accuracy of 97.5%, alongside superior precision, recall, and F1-Scores. These results demonstrate the system’s resilience to environmental variability and real-time applicability in high-security applications. By integrating VIS and IR modalities, this work establishes a scalable and effective framework, advancing the state of biometric authentication technologies. This system represents a significant advancement in multimodal biometrics, paving the way for scalable integration into diverse real-world applications such as border security, healthcare, and financial systems.