Iris identification based on deep convolutional neural networks

With the prominent need for security and the increasing request for information security, security regulations, and trustworthy methods of identification and adjustment in biometric systems worldwide, biometric identification technology transpires extensively used in our everyday lives. With the advancement of science and technology iris biometrics analogous to many further biometric systems offers an alternative solution to this lingering matter. Although iris identification has been immensely studied, it is nevertheless, not a fully solved problem, which is the factor inhibiting its implementation in real-world situations today. The existing iris biometrics technology and other biometrics systems' main problems faced a lack of robustness of the algorithms. But lately, iris identification for biometrics become vigorous by using deep learning. On account of this consideration, iris biometrics technology has gained more attention and interest due to its ability and reliability to dominate a sum of significant limitations of unimodal biometric systems. In this thesis, a new biometric identification is proposed, which is based on a deep learning algorithm (CNN) for identifying humans using biometric modalities of the iris. In the interpretation of the countless performances of deep learning methods in several identification tasks, the structure of the proposed biometric identification is based on convolutional neural networks (CNNs) which segment, extract features, and classify images using SoftMax and ReLu activation function classifiers. Furthermore. VGG-16, VGG-19, ResNet-50 vs ResNet101, MobileNet, AlexNet, and InceptionV3were performed for feature extraction, which attained around 97.68% accuracy on the training data set and almost 98.35% accuracy on the test data set. In iris classification, I have used CNN, ImageNet, and TinyVGG which obtained 98.11% accuracy on the training data set and 98.35% on the test data set. As a highly accurate modality of biometric identification, iris identification uses plenty of mathematical pattern recognition techniques on images of one or both of the irises of a person's eyes.