Systematic analysis of liveness detection methods in biometrie security systems

Biometrie systems suffer from a fundamental vulnerability to presentation attacks which impairs their security and robustness. To overcome this issue, liveness detection methods have been introduced to ensure that input data is originating from a live human being, and it is not forged from old or artificial data. In case of fingerprint and face image-Two of the most widely used traits in biometric systems-liveness detection methods attempts to detect fake input data such as gummy finger or finger image in the former, and printed face image or masks in the latter. A new trend in liveness detection methods is moving toward using bio-electrical signals such as Electrocardiography (ECG) and Electroencephalography (EEG). These signals are believed to have an intrinsic liveness property due to the point that they can only be obtained when a live human subject is present. However this is not a realistic assumption, since in recent works these signals have been artificially crafted, and gotten accepted by biometric systems. Therefore, similar to other traits, liveness detection methods for bio-electrical signals need to be developed, and, furthermore new approaches to liveness detection should be studied. In this paper, we discuss liveness detection challenges, and trade-offs with respect to factors such as usability, cost, and performance, and necessary requirements for an evaluation criteria is specified. Finally, the current state of art in liveness detection methods are comprehensively evaluated using the developed criteria.