Volume - 13 | Issue-1
Volume - 13 | Issue-1
Volume - 13 | Issue-1
Volume - 13 | Issue-1
Volume - 13 | Issue-1
One of the main issues of the present is security. In many facets of modern life, identity verification has grown in importance. It has been demonstrated that hand-based bio- metric features are simple to obtain during data collecting. An increasingly reliable approach of automated personal identifi- cation is finger vein biometrics. Due to the physical traits and properties of the vein patterns in the human finger, which are very impossible to forge, finger vein is a special physiological biometric for identifying individuals. Deep hierarchically taught models (like CNN) have recently outperformed other computer vision algorithms in a variety of applications, but biometrics has received less attention up to this point. This is a key concern because there aren’t enough examples available in biometrics to effectively train CNN. However, because to the enormous number of parameters that the learning algorithm must optimise, deep learning frequently needs a lot of training data. In this study, we provide a novel method of finger-vein image authentication. In order to support our concept, a publicly available vein image data set has been used as a case study. We found that domain-specific and highly discriminative vascular characteristics are provided by transformations learned from such a network. We use basic convolutional neural network (CNN) with transfer learning. The model has been pre-trained using the VGG16 architecture on a variety of image types from the UCI data world. The input from several biometric features is combined by multimodal systems, which not only improve system performance by making up for the shortcomings of each individual trait but also protect the system against attacks like presentation attacks. Moreover, Vein patterns are a powerful biometric identifier due to the uniqueness of blood vessel networks across individuals and difficulties in their reproduction.