On Data-Driven Approaches for Presentation Attack Detection in Iris Recognition Systems

With the development of modern machine learning-based techniques for accurate and efficient classification, the paradigm has shifted to automatic intelligent-based methods. The iris recognition systems constitute one of the most reliable human authentication infrastructures in contemporary computing applications. However, the vulnerability of these systems is a major challenge due to a variety of presentation attacks which degrades their reliability when adopted in real-life applications. Hence, to combat the iris presentation attacks, an additional process called as presentation attack detection mechanism is integrated within the iris recognition systems. In this paper, a review of the modern intelligent approaches for iris presentation attack detection (PAD) mechanisms is presented with a special focus on the data-driven approaches. The presented study shows that the machine learning-based approaches provides better classification accuracy as compared to conventional iris PAD techniques. However, one of the open research challenge is to design the robust intelligent iris PAD frameworks with cross-sensor and cross-database testing capabilities.