Hybrid Image Enhancement and Dense Segmentation for Robust Iris Recognition Using Attention Mechanism

Abstract

Iris recognition is a highly precise biometric identification technique that exploits the unique epigenetic patterns present in the human iris. However, existing approaches often encounter challenges related to segmentation accuracy and classification efficiency. To address these limitations, this research proposes a novel iris recognition framework that emphasizes efficient segmentation and classification by integrating Convolutional Neural Networks with Sheaf Attention Networks (CSAN). The primary objective of this work is to develop an integrated framework that jointly optimizes iris segmentation and classification performance. A Dense Extreme Inception Multipath Guided Upsampling Network is employed to achieve accurate iris segmentation. Subsequently, classifiers, including convolutional neural networks enhanced with sheaf attention mechanisms, are evaluated for recognition performance. Experimental results demonstrate that the proposed approach delivers superior accuracy and robustness in iris recognition, making it well suited for secure authentication and access control applications. Comparative analysis with existing methods shows that CSAN achieves accuracy rates of 99.98%, 99.35%, 99.45%, and 99.65% across four different datasets, respectively.