Biometrics have recently gained significant attention, and have already moved from the lab to commercial applications, mainly in the field of access control and authentication. Traditional biometrics, including iris, face, speech and fingerprints have proven their recognition (determining the identity) and authentication (confirming the identity) potential and are now widely used. However, they have some significant drawbacks since they are obtrusive and uncomfortable for the user, and static physical characteristics can be digitally duplicated (e.g. the face could be copied using a photograph, a voice print using a voice recording, and the fingerprint using various forging methods). Recent technologies in biometrics present more natural ways of recognizing people. Similar to the methods or techniques humans utilize in order to recognize each other, modern trends in biometrics focus on the recognition of dynamic face grimaces, gait, movements, etc. In other words, they tend to recognize liveness rather than static features as the aforementioned traits do. In this respect, behavioral biometrics are related to specific actions and the way that each person executes them.
Recently, the Image and Vision Computing Journal of Elsevier has published a virtual special issue celebrating the breadth of Biometrics that presented the most recent and promising technologies on biometrics. One of the most promising, among others, is the idea of using activity related data for human identification and/or authentication. Given that most of the activities performed in every day life include the human's physical interaction either with other people or with objects, this work attempts to detect and to evaluate a series of stable, invariant, time lasting and unique activity related biometric characteristics for each human. The focus is on the arm's movement and on the movement of the fingers, while performing reaching and grasping activities. The activities that were studied were reaching and grasping an object on a table and reaching and grasping a telephone. The study has demonstrated that very high authentication rates can be achieved using as input only the motion trajectories of the hands and the head.
This result comes to verify recent advances in gait recognition. Gait is another activity similar to reaching and grasping that has been used for recognition and authentication for many years. The same core group of researchers has demonstrated that gait recognition and authentication can be highly efficient in structured environments, where specific conditions on the subject’s gait can be assumed, like walking in corridors, etc.
Behavioral activity-related biometrics like gait or reaching and grasping can lead to identification and authentication systems that are far less obtrusive than those utilizing traditional hard biometrics like iris, or fingerprint. This comes however at a cost of a significantly lower discrimination potential. While traditional biometrics can reach authentication rates up to 99.9%, behavioral biometrics can, usually and based on today’s technology, hardly reach 95%. Therefore, end-to-end applications can use hybrid architectures for their biometric systems including both traditional and behavioral biometrics. For example, access control can be based on a traditional biometric trait, so as to assure high recognition rates. Then, the subject can be continuously monitored unobtrusively using a behavioral biometric trait. This way, potential spoofing attacks of the traditional biometric traits will be discovered by the behavioral biometric system, while the subjects will not have to be distracted from their work for authenticating themselves.
Finally, it can be concluded that even if currently behavioral biometrics can be mainly used in parallel with a traditional biometric trait like iris or face, biometrics technology is expected to advance even more in the near future leading to systems that will be able to perform unobtrusive biometric recognition and authentication.
Top image: Solutions for Society Biometrics - Creative Commons