Iris presentation attack detection: Research trends, challenges, and future directions

Abstract

Currently, interest in biometrics has increased, and personal identity verification is ubiquitous. Iris recognition techniques have recently attracted considerable attention from researchers and are considered one of the most popular topics as they are used for verification purposes. Because of the increasing use of iris recognition, many potential risks have emerged as a natural result of the increased deployment of these technologies. One of the most serious risks is the so-called presentation attack (PA). A PA is the presentation of a sample to an iris sensor to trick the biometric system into making an incorrect decision. Iris presentation attacks are used to spoof or disguise a person’s identity. Many studies have focused on iris presentation attack detection techniques, which are a subset biometric recognition. However, some gaps remain unsolved, and new challenges are rapidly emerging. Despite significant advances in the literature, the problems in iris presentation attack detection have not been adequately addressed and remain open questions. This paper provides a comprehensive overview of iris presentation attack detection from various aspects (e.g., detection techniques, attack types, datasets, and performance measurements). It also attempts to explore the main challenges that may affect presentation attack detection models in terms of important aspects. The challenges that remain to be unresolved are summarised to facilitate problem solving. This review concludes with some directions for future research to help researchers focus on important aspects of the field and try to improve what previous researchers have started. Furthermore, it is likely that this review will be used as a reference for scientists/researchers in the existing science of iris presentation attack detection.