Technological Advances in Biometrics on a Global Scale
Part Two

Basic characteristics of biometric analysis

A good biometric has two basic characteristics: stability and distinctiveness. A stable biometric doesn't change over time. Clearly, hair length would not be a good biometric identifier. A distinctive biometric is unique to an individual. Iris patterns fulfill these requirements; as well as, to varying degrees, fingerprints, face shapes, hand geometries, voices, and signatures. Some biometrics, like fingerprints, may be stable and distinctive but can be temporarily damaged.

No biometric is perfect and the one chosen for a particular use depends on a variety of practical considerations. For ATMs, rapid identity checking is essential, but for employees at a nuclear facility, for example, a lengthier and more intrusive process would be acceptable.

Biometrics fall roughly into one of two categories: physiological and behavioral. Fingerprinting, used widely in forensics and in government databases, is the most well-known physiological biometric. Finger-scanning devices avoid the messy ink associated with fingerprinting by having users touch a glass plate or silicon chip, which records an image of the fingertip's ridges and valleys.

Face recognition is being tapped for applications such as controlling access to a personal computer. A digital camera can take a picture of the person sitting in front of the screen, and comparison software can make sure it matches the owner.

Voice recognition technology is the obvious choice for phone-based systems. Experts in the field disagree over whether voice is a physiological or a behavioral biometric, but it does contain a behavioral component, since a person's voice changes with mood. Voice recognition is one of the cheapest technologies because it uses existing phone equipment, but it is less accurate than other methods.

Dynamic signature verification, a behavioral biometric, assesses the style and speed with which a person signs his or her name. Signatures are already widely used to certify identity, which makes this one of the least intrusive biometrics; however, it is also one of the least accurate.

Requiring an exact match, even if possible, might be counterproductive. Unlike a PIN or a password, the performance of a biometric technology depends a lot on the person using it. Growing a beard or wearing a different type of makeup could trip up a face recognition system. Even pressing a finger down on a glass plate too hard could alter a fingerprint, causing the device to cry impostor.

Biometrics also bring up legal and privacy issues that will have to be addressed as these technologies come into popular use. Consumers already concerned about widespread distribution of personal information are likely to be suspicious of data banks that record encoded details of their body parts.

Much of the wariness may come from the strangeness of a new technology. If biometrics do find their way into computers, automobiles, and cash machines, perhaps their familiarity will also breed less apprehension.