Enhanced Iris Recognition Method by Generative Adversarial Network-Based Image Reconstruction

Iris recognition is one of the non-contact biometric identification methods that are hygienic and highly accurate. involves using iris images obtained by a near-infrared (NIR) camera or visible light camera. A clear image can be when an NIR used, but it requires illuminator in addition to performed with built-in device which also has advantage obtaining three-channel containing color information. Accordingly, studies being conducted on from face taken high-resolution smartphones. However, have unconstrained conditions without cooperation subjects, quality reduced noises such as optical motion blur, off-angle view, specular reflection (SR), other artifacts, thus ultimately deteriorating performance. Therefore, this study, method been proposed for enhancing blurring region deep-learning-based deblurring. In addition, we propose improving performance integrating score periocular regions support vector machine (SVM). The was experimented noisy challenge evaluation-part II training database MICHE database, exhibited improved compared state-of-the-art methods.