Eye Movements Could ID Computer Users


The password is becoming passé. Every week it seems, someone is proposing a new method for authenticating computer users when they attempt to access email or a bank account.

There are finger

swipes, brain

patterns, palm prints,

and even butt-prints.

Now we can add eye movements to the list.


Komogortsev, an assistant professor in the computer science department at Texas

State University-San Marcos thinks that how a person's eyes focus when he looks at an object can be as individual as a fingerprint. 

Google Demos 'Glass' With Crazy Skydiving Stunt

DNEWS VIDEO: Science and Technology Help Find True Love

Eyes make two kinds of movements. One is called a saccade,

which is a swift, jerky motion that results when a person focuses. The other is called a fixation, which is when the eye rests on a given point. Our brains

edit out the saccades most of the time, giving the illusion of continuous

vision, but if you look closely at another person's eyes it's possible to see

the saccades.

Komogortsev designed a system that tracks the eyes, using

several different kinds of stimuli — reading, looking for a dot on a screen and

the classic Rorschach ink blots. With enough data, a computer can build up a

picture of how an individual's eyes move.

Write WIth Your Eyes

The technology is still in its infancy. One problem is that it recognizes the wrong person about one time in three, which is nowhere near good enough.

But it could be combined with

are other types of biometric authentication that rely on the eyes, such as retinal scans, which analyze the pattern of blood vessels on the

retina or iris scans, which analyze the pattern in the iris. Neither of these technologies are fool-proof. Retinal patterns can change due to certain diseases or age, and iris

scans can be fooled by high-quality pictures of the eye.

When combined with Komogortsev's eye movement solution, the error rate for iris scans drops to about 5 percent, which is

more in line with other technologies.

Via TechNewsDaily

Credit: Wikimedia

Commons / Jake Maheu

Invalid Email