An Efficient and Robust Iris Segmentation Algorithm Using Deep Learning

Abstract

The iris segmentation algorithm is very essential in an absolute iris recognition system and has a direct influence on the verification and recognition results of the iris. However, traditional iris segmentation algorithms have poor adaptability and are not robust enough when used in noisy iris databases captured under infinite conditions. Under non-ideal conditions, existing segmentation based on local operations cannot see the true iris boundary, and iris recognition result come in failure. In addition, paper discuss on the iris distribution algorithm that shows the benefits from the convolution neural network (CNN).