AI-Driven Multimodal Biometric Classification: Improving Recognition Accuracy Using Finger, Face, and Ear Biometrics

Biometric recognition has emerged as a critical component of secure identity verification systems. While unimodal biometrics such as fingerprint, face, or ear recognition have been widely researched, they suffer from limitations related to noise, occlusion, and spoofing. This paper proposes an AI-driven multimodal biometric system integrating fingerprint, face, and ear modalities to enhance recognition accuracy and robustness. Using Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) for feature extraction and a fusion-based classification strategy, the proposed approach is conceptually shown to outperform unimodal systems. A literature comparison and expected results suggest that the fusion model can achieve recognition accuracy of approximately 97–98%, surpassing most existing methods. The study concludes by highlighting the potential of multimodal biometrics for real-world applications in high-security domains.