Iris Recognition System (IRS) In Biometric World

Abstract

With the rapid growth of digital technologies, the demand for reliable and secure authentication methods has become increasingly vital. Among various biometric identification techniques, iris recognition stands out as one of the most dependable approaches due to the iris’s unique and stable pattern that remains consistent throughout a person’s life. Unlike traditional password- or token-based systems, iris recognition offers a higher level of accuracy, security, and resistance to environmental and age-related variations.  This paper presents a comprehensive study of the Iris Recognition System (IRS), encompassing both classical and modern methodologies. The research highlights the traditional techniques such as Daugman’s rubber sheet model and Gabor wavelet-based feature extraction, while addressing their limitations in uncontrolled or noisy conditions. To overcome these challenges, the proposed approach integrates advanced deep learning methods like Convolutional Neural Networks (CNNs) and transfer learning models such as DenseNet201, which enhance feature extraction and classification accuracy. The workflow includes essential stages such as image acquisition, preprocessing, iris localization, normalization, feature extraction, and pattern matching. The proposed system was evaluated on standard datasets including CASIA, UBIRIS.v2, and IITD, achieving a recognition accuracy of over 95% even under challenging conditions like poor lighting and off-angle captures. The paper also explores key applications of iris recognition in domains such as border control, mobile authentication, e-governance, and healthcare. Finally, it discusses potential advancements like lightweight CNN architectures, multimodal biometric systems, and edge-based iris recognition models for future research and deployment.