#biometric methods

Voice Scanning Voice scanning and authentication systems create a voiceprint from a speech sample and compare it to the voice of the person being verified. It is a quick and nonintrusive technique that is mainly useful for remote transactions such as telephone access to banking information. Behavioral Biometrics Biometrics are essentially invariant patterns, and these can be found in behavior as well as in physical features. One of the most hopeful techniques analyzes the pace or rhythm of a person’s typing on a keyboard and generates a unique numeric code. An analogous approach might be applicable to mouse usage.

Facial scanning uses cameras and image analysis software that looks at areas of the human face that change just a little during the course of life and are not simply alterable, such as the upper outline of the eye sockets and the shape of the cheekbones. Researchers at MIT developed a series of about 125 grayscale images called eigenfaces from which features can be united to differentiate any given face. The template resulting from a scan can be compared with the one on file for the claimed identity, and coefficients expressing the level of similarity are calculated. Variance above a specified level results in the person being rejected. Facial scanning is often viewed as less invasive than the use of fingerprints, and it can also be applied to surveillance images.

These techniques take benefit of many unique individual characteristics of these parts of the eye. The scanned characteristics are turned into a numeric code similar to a bar code. Retina scanning can be uncomfortable because it involves shining a bright light into the back of the eye, and has usually been used only in high-security installations. However, iris scanning involves the front of the eye and is much less disturbing, and the person being scanned needs only to look into a camera.

This technique measures a number of characteristics of the hand, including the height of and distance between the fingers and the shape of the knuckles. The person being scanned places the hand on the scanner’s surface, aligning the fingers to five pegs. Hand-scanning is practically accurate in verifying an individual compared to the pattern on file, but not accurate enough to identify a scan from an unidentified person.

Finger scanning involves the imaging and automatic analysis of the pattern of ridges on one or more fingertips. Unlike traditional fingerprinting, the actual fingerprint is not saved, but only enough key features are retained to provide a unique identification. This information can be stored in a database and also compared with full fingerprints stored in existing databases (such as that maintained by the Federal Bureau of Investigation). Finger scanning can meet with resistance because of its similarity to fingerprinting and the association of the latter with criminality.