Hierarchical Convolutional Neural Network based Iris Segmentation and Recognition System for Biometric Authentication
Elavarasi Gunasekaran, Vanitha Muthuraman · 2020
Multimodal biometric models are commonly employed in various real time applications because of its capability of dealing with several noteworthy constraints like noise sensitivity, intra-class variability, and susceptibility to spoofing. This study presents a new real time bio-metric model based on Deep Learning (DL) architecture called Hierarchical Convolutional Neural Network (HCNN). The proposed model involves several stages. At the initial level, the boundaries are attained from the given image for providing Region of Interest (RoI) to the subsequent level. Next, the resulted image comprises the regions of upper as well as lower eyelids and remaining regions like skin, eyelashes, and sclera. At the next level, inside the RoI, HCNN is employed for providing the boundaries of actual iris with the help of learned features. A detailed experimental analysis takes place to ensure the efficacy of the projected technique. The obtained simulation outcome verified the superior characteristics of the presented model.