Iris Liveness Detection: Current Approaches and Future Prospects

This paper focuses on the role of liveness detection in protecting iris recognition systems from spoofing attacks that may use artificial objects such as printed images, contact lenses, or synthetic irises. This review covers the current state of the art in liveness detection methods, starting from the traditional handcrafted features like Local Binary Patterns to the deep learning models such as VGG-Net. We discussed commonly used datasets which include CASIA-Iris, Clarkson LivDet-Iris, and ND-CrossSensor-Iris. These datasets also pose some challenges when trying to achieve real world applicability. Some of the major issues that have been identified including the ability of the algorithms to adapt and generalize, the diversity of the datasets and the computational power requirements. Further, the future direction comprises developing more diverse datasets, lightweight models for mobile application and multimodal biometric systems to improve the robustness against the ever-evolving spoofing attacks. This review also emphasizes on the need for the constant evolution of the iris liveness detection to improve the security of the biometric systems.