Key points are not available for this paper at this time.
Abstract Biometric systems authenticate individuals based on unique physiological or behavioural traits for identification and access control, providing heightened security compared to traditional password-based methods. Automatic fingerprint recognition is widely employed for its reliability and uniqueness, yet fingerprint classification remains challenging in large-scale recognition systems due to variability, noise, and ambiguity. This paper presents a deep learning approach to fingerprint classification, utilizing the VGG16 model for feature extraction and classification. By employing precise algorithms, our method accelerates identification by reducing the number of comparisons needed during fingerprint retrieval. Our framework, tested on the FVC2004 DB1 database, demonstrates significant advancements in accuracy, highlighting deep learning's efficacy in enhancing biometric identification systems.
Prateek Nahar (Wed,) studied this question.