Iris Liveness Detection using Deep Learning Networks

A specific biometric characteristic must be unique for each person for whom it can be determined and invariant over time in order to be used for identification purposes. Biometrics with notable limitations includes signatures, images, fingerprints, voiceprints, as well as retinal blood vessel patterns. Even though they are inexpensive, simple to produce, and convenient to retain, signatures and pictures cannot be reliably identified mechanically and are easily falsified. The individual iris, in contrast side, is an ideal biometric verification for identifying with ease, speed, accuracy, and automation because it is a vital eye organ and is well protected from the surrounding environment while also being clearly visible from within one metre of range. Iris recognition is the most reliable and precise biometric method in the current trend. Biometric authentication is an automated biometric identification technology that analyses photographs of a person's eye irises to identify their distinctive, complicated random patterns using mathematical pattern recognition techniques. The existing work uses the knowledge distillation method for the extraction of the features of the periocular regions. In this work, it is suggested to develop a face and iris identification system, where the face, eye, and iris region are segmented using the Grassmann method, Curvelet transform, and deep neural networks. Real-time enrolment system features are used to construct a template of the identified region utilising template matching for recognition. The outcomes demonstrate that the suggested approach is effective for current iris image-based biometric recognition with improved accuracy.