Iris Recognition Using A Deep Learning Approach

Abstract

Iris recognition has been an lively research area during last few periods, because of its wide bids in security, from aerodromes to home security border control. Diverse structures and procedures have been projected for iris recognition in the ancient period. In this paper, we suggest an endwise deep learning framework for iris recognition based  on convolutional neural network (CNN), which can mutually acquire the feature  illustration  and  achieve  acknowledgment.  We train our model on a well-known iris recognition dataset using only a few training images  from  each  class,  and  show  hopeful  results and  enhancements  over  previous  approaches.  We  also  present a  visualization  technique  which  is  able  to  perceive  the  important areas  in  iris  images  which  can  mostly  influence  the  recognition results. We trust this framework can be commonly used for other biometrics recognition errands, helping to have a more ascendable and exact systems.