Iris Liveness Detection Using a Cascade of Dedicated Deep Learning Networks

Iris pattern recognition has significantly improved biometric authentication due to its high stability and uniqueness. Such physical characteristics have been essential in security applications and other related areas. However, presentation attacks, also known as spoofing techniques, can bypass biometric authentication systems using artefacts such as printed images, artificial eyes, textured contact lenses, and others. Today, many liveness detection methods that improve the robustness of these systems have been proposed. An ideal Presentation Attack Detection (PAD) technique should be able to detect all these attacks, along with any new or unknown presentation attack instrument species that may be developed in the future. PAD for iris recognition systems is a very dynamic topic, as shown in past editions of the LivDet competition, revealing that there are still open problems to get efficient methods for capturing devices. This work contributes to improving the state of the art, adds a new database and also explains the methodology used for our team.