Open AccessIris Recognition using Convolutional Neural Network Design
Author(s) -
Gajanan Choudhari,
Rajesh Mehra
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.i1108.0789s19
Subject(s) - softmax function , computer science , artificial intelligence , convolutional neural network , pattern recognition (psychology) , generalization , classifier (uml) , iris recognition , backpropagation , feature extraction , feature (linguistics) , artificial neural network , machine learning , mathematics , mathematical analysis , linguistics , philosophy , biometrics
Iris trait has gained the attention of many researchers recently as it consists of unique and highly random patterns. Many methods have been proposed for feature extraction and classification for iris trait but suffer from poor generalization ability. In this paper, a scratch convolutional neural network is designed in order to extract the iris features and softmax classifier is used for multiclass classification. The various optimization techniques with backpropagation algorithm are used for weight updating. The results show that the Convolutional Neural Network based feature extraction has proven to provide good generalization ability with improved recognition rate. The effect of various optimization techniques for generalization ability is also observed. The method is tested on IITD and CASIA-Iris-V3 database. The recognition rates obtained are comparable with state of art methods.