DEEP CONVOLUTIONAL NEURAL NETWORK-BASED BIOMETRIC AUTHENTICATION WITH MINUTIAE EXTRACTION FOR ENHANCED LICENSE VERIFICATION SYSTEM

Driving license systems are very difficult to monitor using conventional methods. The conventional verification system suffers from high time complexity and chances of corruption. The fingerprint is one of the best biometric features and can be effectively utilized for the verification of a license. Here, the method based on Deep Convolutional Neural Network (DCNN) is devised for license verification. Here, the various entities, such as users, traffic cops, the Road Transport Department (RTO), blockchain, and cloud servers, are considered. Authentication is performed here at two levels, with the first level processed through the user, RTO, and state server, and the second level carried out between the state and central servers. License verification is effectuated using the fingerprint image, which is first subjected to minutiae extraction. Here, the minutiae extraction is performed using DCNN and the extracted minutiae and minutiae locations are saved in the database. A similar process is performed with the query fingerprint image, and the minutiae location is found. Both minutiae locations are matched using a hybrid similarity measure formulated using the Harmonic mean and Matusita distance. Besides, the devised technique provides better output with a False Acceptance Rate (FAR) of 12.90%, False Rejection Rate (FRR) of 9.03% and Equal Error Rate (EER) of 13.22%.