Artificial Neural Network-Based Fingerprint Classification and Recognition

The most commonly used biometric technique for identifying people is fingerprint-based biometrics. It is divided into two parts: verification (if this individual is genuinely himself) and identification (identifying a person from a pool of persons). Due to the enormous number of comparisons required, the Automatic Fingerprint Identification System (AFIS), which typically conducts two stages: feature extraction and matching, had difficulties with a large database of fingerprint photos for the real-time application. So, more classification stages for complete fingerprint data can make it faster for the AFIS to identify a person. In this paper, we presented a classification method for identifying detailed fingerprint information by utilizing a deep learning approach to support the operations for classifying, identifying, and recognising the fingerprint. The proposed method was designed to differentiate certain fingerprint information, such as left-right hand classification, sweat-pore classification, scratch classification, and finger classification. We privately created our fingerprint image dataset due to high personalization and security concerns (25 fingerprint images in the dataset with seven features for each image through the scanning technique). Finally, the research results for the proposed study were accurate and outperformed previous results.