Finger vein based biometrics identification

Abstract

Form past few decades’ biometrics has gained a whole lot of attention due to its enormous benefits in the security field. Like every other system, the biometric identification systems also have their limitations and drawbacks, yet the best trait amongst all of them so far is vascular biometric. In our project, the subject of interest are finger veins which are captured using infrared imaging system. Since veins are internal traits so they can neither be contaminated nor captured without user willingness. This project is implemented using Convolutional Neural Network (CNN) but a typical problem with CNN is that after training we cannot add new subject to the dataset. To deal with that problem we have also implemented our system using Siamese network. In short, this project is implemented using two techniques named as CNN and Siamese Network