Biometrics come of age

Toby Howard

How do you prove you're you? Most personal identification systems are low-tech: a signature, a passport photograph, a memorised password or PIN number. Soon, these'll be obsolete. "Biometric" recognition is now set to go mainstream, bringing a host of hi-tech ways of finding out if you really are who you say you are.

Evolving from decades of scattered research, biometrics – measuring your body's characteristics to identify you uniquely – is now recognised as a science in its own right, an achievement highlighted by the publication last year of the first-ever book devoted to the subject ("Biometrics: Personal Identification in Networked Society").

The oldest biometric system is fingerprinting, in use since the late 19th century. Now you can get automated fingerprint systems for as little as $150 from www.digitalpersona.com

If you buy a season ticket to Disney World, you'll only get in if the shape of your hand matches "your" hand profile in a database. Several other access-control systems now use face recognition (see, for example, www.visionics.com).

Iris scanning is another proven method, which works by extracting patterns from a video-image of the eye, encoding features such as the colour of the iris, the edge patterns of the pupil, and the reflection of the light from the eye itself. Iris scanning was used for athlete security at the 1998 Winter Olympics, and several US banks are currently evaluating the technique for staff security.

Like your eyes, the shape of your ears can say a lot about you. In fact ear prints are now a recognised identification tool in forensic science, and in 1998 a man was sentenced to life imprisonment for murder on the strength of an ear-print he left on a window at the crime scene.

Other methods include analysing your keyboard dynamics – your typing speed and how often you make your own repertoire of typing errors – and using infra-red photography to image the vein patterns in your retina, face, wrist, or hand.

And there are also some slightly unusual methods in development. Computer vision researchers at Southampton University, for example, are trying to recognise people by the way they walk. Traditionally, human and animal gait analysis has relied on fixing markers at strategic positions on the limbs being studied. The changing positions of the markers as the subject walks are easy to record, and also allow the actual limb structure to be replaced by a simplified "stick figure". For biometric purposes, however, you want to work from video images of people walking, without any markers attached, and this is hard to do. First you have to process each video frame to recognise the legs as distinct from background and foreground features; then you have to track their movement between frames and find a way to characterise the motion so that it can be compared against a database. Recognising that this is a "fuzzy" problem, the Southampton team are working on a statistical solution, but don't have one yet.

Then there's odour measurement. Several companies are developing systems that will identify you by your body odour (don't worry, they normally choose to sniff the skin on your hand). Apparently the particular blend of smells cooked up by a human body is unique to each person. But measuring and analysing body odour is hard, since most of us go to extraordinary lengths to mask any trace of it.

Whatever the method, biometrics is coming – like it or not. In the future you're going to find it very hard not to be you.

Toby Howard teaches at the University of Manchester.