A Supervised Learning Model Design to Evaluate Palmprint Images for Biometric Authentication using Artificial Intelligence

Biometric authentication has gained significant traction as a reliable and secure method for identity verification. This study introduces a novel model, PalmSecureNet, for palmprint-based biometric authentication. Leveraging the Sapienza University Mobile Palmprint Database (SMPD) comprising 3677 images, the proposed model employs advanced preprocessing techniques and an optimized convolutional architecture. PalmSecureNet incorporates depth-wise separable convolutions and residual connections, ensuring efficient and accurate feature extraction. The model evaluated against nine state-of-the-art methods, achieving a remarkable accuracy of 98.21%, outperforming all baseline models. The comprehensive evaluation included scenarios with varying lighting conditions, noise levels, and computational constraints. Results demonstrated that PalmSecureNet excelled in robustness and efficiency, making it suitable for real-time applications. This work highlights the potential of PalmSecureNet as a dependable solution for biometric authentication, addressing challenges like environmental variability and resource limitations.