Iris Location and Recognition by Deep-Learning Networks Based Design for Biometric Authorization
Cheng-Shun Hsiao, Chih‐Peng Fan, Yin‐Tsung Hwang · 2021 IEEE 3rd Global Conference on Life Sciences and Technologies (LifeTech) · 2021
In this work, the effective deep-learning network based design is studied for iris biometric authentication. Firstly, the system locates and detects the position of the iris by the YOLO based deep-learning detector, and then it strengthens the iris features by histogram equalization. Finally, the iris image is classified by the VGG-16 based model. By the self-made near infrared (NIR) database, the proposed deep-learning based design achieves the recognition accuracies up to 98% at the mode with intruders. Besides the performances of false acceptance rate (FAR) and false rejection rate (FRR) are also evaluated. Compared with the previous studies, the proposed design provides the effective performance of accuracy and does not need the iris segmentation process.