CNN Based Iris Recognition System: A Novel Approach

8 Pages Posted: 30 Nov 2020

See all articles by Sheena S

Sheena S

Cochin University of Science & Technology, School of |Engineering

Sheena Mathew

Cochin University of Science & Technology, School of Engineering

Date Written: November 23, 2020

Abstract

Iris image is the most important and unique feature in the field of biometrics and a powerful tool for human identification because of its strong and unique textual information. Extracting these prominent features is the main advancement in recognizing the iris biometric modality. The problems we faced in computer vision can be overcome with the success in deep learning technology. The features learned using CNN was found to be very much useful in the Iris recognition system. This work mainly focused on evaluating the extracted features learned from the proposed convolutional neural network model and Softmax Classifier. The selective features of the source image are extracted by this Softmax Classifier without any preliminary domain knowledge. The input image represents the localized, normalized, and then enhanced iris region, then it will be classified into N classes. The performance of the system has been experimented on, with two public datasets viz IITD and CASIA. An accuracy of 96 % is obtained by using this proposed system.

Keywords: Iris Recognition; Biometric system; CNN; Deep learning

Suggested Citation

S, Sheena and Mathew, Sheena, CNN Based Iris Recognition System: A Novel Approach (November 23, 2020). Proceedings of the 2nd International Conference on IoT, Social, Mobile, Analytics & Cloud in Computational Vision & Bio-Engineering (ISMAC-CVB 2020), Available at SSRN: https://ssrn.com/abstract=3735924 or http://dx.doi.org/10.2139/ssrn.3735924

Sheena S (Contact Author)

Cochin University of Science & Technology, School of |Engineering ( email )

Sheena Mathew

Cochin University of Science & Technology, School of Engineering

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
569
Abstract Views
1,438
Rank
88,561
PlumX Metrics