Identity Theft and Biometrics

In this high-tech world in which we live, there are constant stories of identity theft and that is an unfortunate thing. Twenty years ago the worst thing that could happen to you is getting your car ripped off… Welcome to 2009: They’ll steal your house too! One way which consumers have sought to protect their identity as well as their valuables is though biometric identification. However, as always it seems that the crooks are only one step behind the cops. Or in this case, high-tech criminals are finding new ways to break the seemingly unbreakable security that biometrics provide. Researchers have already shown that fingerprint identification security measures can be spoofed using a fake finger made of gelatin. Now while identity theft is bad, your credit card number, social security number or drivers license can be changed if they are compromised. So… if someone steals your fingerprint, can you change that? So is there anything we can do to make hijacking your fingerprint more difficult? Luckily there is!! One such approach is “liveness testing”. Researchers have been coming up with some unique ways to determine if the fingerprint being presented to the biometric system is a real finger, attached to a living human being (don’t laugh! We all saw the movie Demolition Man… remember the Iris authentication scene?? hehehe) Another way to mitigate the theft of your biometric data is to keep as few copies of the real data as possible. In other words for, if I wanted to recognize you I could keep a photo of you on hand and that photo can be stolen, but there are ways of storing a representation of what you look like without keeping the image. So for example, instead of keeping your image I could keep a list of descriptive features about you: brown hair, brown eyes, nose being located 17mm above the upper lip, eyes being spaced X millimeters apart. This sort of “feature vector” representation is already in use, but again… there is some research going on in Italy where scientists have created an algorithm that recreates a fingerprint from the feature vector that could match on a machine matcher.