biometrics

The earliest use of biometrics was probably the development by Alphonse Bertillon in 1882 of anthropometry, a system of classification by physical measurements and description. While this was soon supplanted by the discovery that fin-gerprints could serve as an easier to use means of unique identification of persons, the need for a less invasive means of physical identification has led to the development of a variety of biometric scanners that take Bertillon’s ideas to a much more detailed level.

Technologies

In general, biometric scanning involves four steps: the capture of an image using a camera or other device, the extraction of key features from the image, the creation of a template that uniquely characterizes the person being scanned, and the matching of the template to stored tem-plates in order to identify the person.

There are several possible targets for biometric scan-ning, including the following areas:

Facial Scanning

Facial scanning uses cameras and image analysis software that looks at areas of the human face that change little during the course of life and are not easily 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 combined to characterize 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 degree of similarity are calculated. Variance above a specified level results in the person being rejected. Facial scanning is often viewed as less intrusive than the use of fingerprints, and it can also be applied to surveil-lance images.

Finger Scanning

Finger scanning involves the imaging and automatic analy-sis 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.

Hand Geometry

This technique measures several 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 reasonably accu-rate in verifying an individual compared to the template on file, but not accurate enough to identify a scan from an unknown person.

Iris and Retina Scanning

These techniques take advantage of many unique individ-ual 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 generally been used only in high-security installations.

However, iris scanning involves the front of the eye and is much less intrusive, and the person being scanned needs only to look into a camera.

Voice Scanning

Voice scanning and verification systems create a “voice-print” from a speech sample and compare it to the voice of the person being verified. It is a quick and nonintrusive technique that is particularly 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 promising techniques (recently patented) analyzes the pace or rhythm of a person’s typing on a keyboard and generates a unique numeric code. A similar approach might be applicable to mouse usage.

Applications of Biometrics

Due to the expense of the equipment and the time involved in scanning, biometrics were originally used primarily in verifying identity for people entering high-security installa-tions. However, the development of faster and less intrusive techniques, combined with the growing need to verify users of banking (ATM) and other networks has led to a growing interest in biometrics. For example, a pilot program in the United Kingdom has used iris scanning to replace the PIN (personal identification number) as a means of verifying ATM users.

The general advantage of biometrics is that it does not rely on cards or other artifacts that can be stolen or other-wise transferred from one person to another, and in turn, a person needing to identify him or herself doesn’t have to worry about forgetting or losing a card. However, while workers at high-security installations can simply be required to submit to biometric scans, citizens and consumers have more choice about whether to accept techniques they may view as uncomfortable, intrusive, or threatening to privacy.

Recent heightened concern about the stealing of per-sonal identification and financial information (see identity theft) may promote greater acceptance of biometric tech-niques. For example, a built-in fingerprint reader (already provided on some laptop computers) could be used to secure access to the hard drive or transmitted to authenti-cate an online banking customer.

Of course every security measure has the potential for circumvention or misuse. Concerns about the stealing and criminal use of biometric data (particularly online) might be addressed by a system created by Emin Martinian of the Mitsubishi Electric Research Laboratories in Cambridge, Massachusetts. The algorithm creates a unique code based on a person’s fingerprint data. The data itself is not stored, and the code cannot be used to re-create it, but only to match against the actual finger.

The growing use of biometrics by government agencies (such as in passports and border crossings) is of concern to privacy advocates and civil libertarians. When com-bined with surveillance cameras and central databases, bio-metrics (such as face analysis and recognition) could aid police in catching criminals or terrorists, but could also be used to strip the anonymity from political protesters. The technology is thus double-edged, with the potential both to enhance the security of personal information and to increase the effectiveness of surveillance.