Integrating ocular and iris descriptors for fake iris image detection

Iris recognition has emerged as one of the most promising contactless biometrics technologies to provide automated human identification. Several national ID programs, such as Aadhar in India, incorporate iris biometrics to provide unique identity to millions of citizens. Therefore it is vital that integrity of such large scale iris deployments must also be safeguarded. Iris recognition technologies are increasingly becoming susceptible to sophisticated sensor level spoof attacks. This paper details the development of a new anti-spoofing approach which exploits the statistical grey-level dependencies in both the localized and global eye regions surrounding iris. We present experimental results on publicly available fake iris image database. The correct classification rate of 99.75% is obtained from the developed spoof iris detection approach using 1200 real and fake iris images and rom a publicly available database.