A Three Steps Eye-Liveness Validation System

Abstract

Iris recognition technique is widely used in different sectors of security and authentication systems because of its effortlessness and high level of safety. However, this type of system is still facing serious challenges. For instance, spoofing of the human iris through printed photos or recorded videos that act as real iris can easily trick the authentication system.In this paper, we address the challenge posed by spoofing and propose a novel approach for eye liveness detection based on three steps. In the first step the person opens his eye and the system detects the changes in pupil size through responding to the ambient illumination. As per the second step, the person closes his eye and the system recognizes the blood streaming signals under the skin of eyelid. In the third step, the person opens his eye again and the system detects the pupil size again and the system compares the results of the pupil size. For the liveness to be validated, two checks have to be passed. The first one is related to the pupil size: If the results of the pupil size in the first step differs from the third step, then the eye pupil could be real because it responded to the ambient illumination, and the second check is invoked—otherwise, liveness is not validated. The second check is related to the detection of heart beat from the eyelid. Should also this second check be passed, the liveliness of the user is assumed, otherwise it is rejected.To test our proposal, we have created two dataset, populated by 18 volunteers. the first dataset is for the pupil sizes, and second one for the heart pulses captured from the eyelid. Experimental results conducted employing the proposed solution do support its quality and viability.