Human Computer Interface Technology

Biometrics Introduction and Issues
October 21, 2002

*Copyright 1999-2002,
Perry R. Cook,
Princeton University


Biometrics is the study and use of measurable biological characteristics.

In HCI, biometrics refers to authentication techniques that rely on measurable physical characteristics that can be automatically checked.

The Biometrics Consortium defines it as "automatically recognizing a person using distinguishing traits (a narrow definition)"

Examples include computer analysis of fingerprints or speech.

One reference collection is available at The International Biometric Group

Biometric Tradeoffs

There are a number of tradeoffs when selecting a biometric. Those include:

Many of these depend greatly on the technology.

Another aspect of BioMetrics is whether they are used for

Identification ("Who Am I")
is a much more difficult task
than the more limited

Verification task
("How Likely is it that I am Who I Say I Am").

A further important aspect of Identification systems is the rate of

and the rate of

Another related issue is whether the population under question (cleint base, audience, whatever) is


Other issues have to do with the training required for a user to be scanned and recognized/verified.

Much of biometrics is simple pattern recognition, classical statistical, neural net, fuzzy, etc. But as with any pattern recognition system, constructing (or buying) the sensors to get the information you want, and extracting the right features from the sensor data, makes all of the difference between a useful system and a useless system.

Fingerprints and Fingerscans

Fingerprints have been in use for a long, long time (may decades) for forensics identification.

Sensor technologies for acquiring the data include Thermal, Capacitance, Ultrasound, and Optical.

Typical features are "minutae," which are the little bumps, breaks, rapid shifts, etc. in the otherwise smooth curves of the fingerprint pattern.

Some Technology and background are located here

Facial Scanning

Of course, humans are very good at extracting information from faces and images of faces. Turns out that machines can, in some cases, get even more information from facial images than humans can. [Golomb, Lawrence, and Sejnowski, 1991, "Sexnet: A neural network identifies sex from human faces." in Advances in Neural Information Processing Systems]

Typical acquisition sensor technology is a cheap camera.

Typical features are measurements of some major facial components. These are usually selected to be those that are not alterable by frowning, talking, smiling, etc. Examples include eye socket extrema, sides of the mouth, cheekbone regions, etc.

Some Technology and background are located here

Hand Scanning

Basic measurements of the human hand make for an inexpensive verification technique (not generally good enough for identification).

Sensor technologies include simple slide pots, capacitance, video (most common).

Features include lengths of fingers, distance between joints, and widths of knuckles.

Some Technology and background are located here

Iris Scan

Iris is considered to be unique (odds of two irises matching within 75% is < 1 in 1016).

Sensor: Camera

Features: trabecular meshwork (radial pattern formed before birth), other rings, freckles, etc.

Some Technology and background are located here

Retina Scan

Of course, this is the "biggie" of spy and science fiction movies. It's also old by biometric standards (1930's observations about uniqueness of retinae). It must be good then, right?

Sensor: Camera

Features: Blood vessel patterns on back of inner eye (retina)

Problems: 1/2" range, trained user, cooperative user, patient user

Some Technology and background are located here

Voice Scan (Speaker Identification/Verification)

Next to retina, this is clearly the sci-fi/spy movie winner. So this must be a solved problem too, right?

Sensor: Microphone

Features: "Qualities of the voice" (lots and lots of these)

Problems: It's hard!!!!

Issues: Text dependent vs. independent
microphone placement/environment
colds, coughs, mimics, modification, etc.

Signature, Handwriting Recognition

Like voice. Not a solved problem, hard, forgeable, ... ... ... ...


Open forum ...

Other Metrics, Future Metrics

Body measurements

Timing of activity (walking gait, etc.)


Brain Activity

. . . . . .

* Permission to make digital or hard copies of part or all
of this work for personal or classroom use is granted with
or without fee provided that copies are not made or dis-
tributed for profit or commercial advantage and that copies
bear this notice and full citation on the first page. To copy
otherwise,to republish, to post on services, or to redistribute
to lists, requires specific permission and/or a fee.

Back to Princeton HCI Page