Personal Information Security Using Biometric Traits

Abstract

to the accelerated development in the fields of Science and technology, traditional means of protecting personal information such as cards and passwords are becoming less effective and more vulnerable to hacking and fraud. This has led to the emergence of biometric identification technologies as effective, secure and easy-to-use solutions, being based on the unique physiological or behavioral characteristics of individuals, making them suitable for many applications such as criminal identification, access control systems, correctional institutions, and banks. In this context, hand pattern recognition is one of the techniques distinguished by its high accuracy and its ability to extract rich characteristics from the hand image. In this work, we propose a biometric recognition system based on deep learning technologies using bypass neural networks (CNN), for which three well-known models were applied: AlexNet, VGG16, and VGG19. The system was evaluated in the cases ofUnimodal and multimodal , with an analysis of the impact of the application of Tied Rank Normalization technology on performance. The simulation results showed the effectiveness of the models used, and also highlighted the differences between the single and multiple systems, which helps in determining the optimal strategy in terms of accuracy and efficiency.