Biometric Authentication Based on Finger Knuckle Print Using Deep Learning Methods

Identifying individuals across various institutions and organizations has become an essential and critical task, particularly in security-sensitive applications where maintaining national security is of paramount importance. Biometric characteristics provide reliable means for personal identification, among which finger knuckle print has attracted considerable attention. This paper focuses specifically on the use of finger knuckle print for biometric authentication. The finger knuckle regions were first extracted and preprocessed from hand images. In fact, after collecting images of the backs of different individuals' hands, the finger knuckle regions were extracted. Subsequently, advanced deep learning techniques were employed to extract discriminative features from the images, enabling accurate authentication. To evaluate the proposed approach, finger knuckle print regions obtained from a hand image dataset collected by the author were utilized as the test benchmark. The experimental results confirm the reliability of the proposed approach, highlighting its potential applicability across various real-world authentication and security systems.