You might think that PIN codes and fingerprints are pretty secure identity systems, but they are in fact simple to hack. The criminal community has found it too easy to steal PIN numbers just using cameras, card copiers or the point of a knife at the cash point. And James Bond famously tricked an adversary to believing his false identity by wearing ‘fake fingerprints’.
The use of biometric identification — using the unique properties and characteristics of an individual to help identify them — continues to grow in popularity. Modern electronic passport checks use face recognition, and iris scanning has also been tested in some airports.
Recognising the growing market for less obtrusive biometric identification, the project ‘Unobtrusive authentication using activity related and soft biometrics’ (Actibio), has been part-funded by the EU to look at whether more dynamic features, such as the way people walk, talk or respond to specific stimuli, could also be used for user verification purposes.
‘We are each so unique,’ remarks Dr. Dimitrios Tzovaras, who coordinates Actibio. But it is not just how you look and your physical features. The way you move and respond, even the pattern and shape of your heartbeat, these all have unique features too. Our project is one of the first to look seriously at these more dynamic features and find ways of spotting those unique characteristics that categorically say that this is you, not someone else.’
The project builds on the findings of its predecessor, the Humabio project, which demonstrated the feasibility of using multimodal behavioural and physiological biometrics for reliable user authentication. Actibio is now using and refining the algorithms developed by the Humabio team and testing them in several real-world workplace and security applications.
‘We have tested our technologies with more than 100 volunteers in several different settings including vehicles and control rooms. So far we are very encouraged with the results,’ remarks Dr. Tzovaras. ‘We have found that the use of these “soft” biometric measures really enhance recognition rate when you combine them with existing biometric systems.’
Dr. Tzovaras says that using the technology alongside face recognition could enhance security in banks and cash points, for example. Faces could be scanned at the counter or the ATM, but the way someone walks up to the cash point could also be scrutinised providing a double check for the facial system.
The project is also testing a special sensing seat or cushion which you can put in the cab of a truck, for example, and use it for extracting so-called ‘anthropometric profiles’ based on the user’s weight distribution on the seat and the way the seat cover deforms. These metrics can then be used to identify the driver. This would make it impossible for the vehicle to be hijacked or stolen.
Of course all these biometric systems must first be ‘trained’, for example to recognise what makes your gait different to everyone else’s. Typically, you would be filmed as you walked in highly controlled conditions. Image analysis software can track your body joints as you move; the Actibio algorithms then look for the distinguishing features in this movement.
‘On the surface the results from our dynamic recognition trials as a technology on its own do not look that impressive. They have an equal error rate of about 3 %, which means that there is a mismatch for about 3 in every 100 people,’ says Dr. Tzovaras, ‘but this is a revolutionary improvement for dynamic recognition. And when we combine dynamic and static biometric systems the equal error rate drops to zero; the identification is correct every time. We can see many excellent applications for authenticating individuals and monitoring their behaviours without having to intrude into or interrupt what they are doing.’
written by
biometrics2013
