OPTICS

  A Discerning Eye

  In the James Bond movie, "Never Say Never Again," a camera zooms up to a
character to identify him by the unique appearance of his eye. At that time,
there was no device that could accomplish such a thing. But now Sensar, a
subsidiary of the David Sarnoff Research Center, has announced a $25.8-million
agreement with OKI Electric Industry Ltd. in Tokyo, one of the world's leading
suppliers of automated teller machines (ATMs). This means iris recognition
could be coming to an ATM near you.

  Unlike signature verification, voice recognition or fingerprinting, iris
recognition requires little cooperation. A person simply walks up to the
machine and inserts his or her bankcard. Meanwhile an ordinary video camera
captures an image of the customer's right or left eye. This image is converted
into a digital code, which is compared with one already stored for that
individual. If the system perceives a match, the customer can proceed. The
process takes about five seconds.

  Although color is the first thing we notice about someone's eyes, recognition
is based only on immutable structures of the iris.  These include the
trabecular meshwork of connective tissue, collagenous stromal fibers, ciliary
processes, contraction furrows, crypts, vasculature, rings, corona, coloration
and freckles. As with fingerprints, most of these characteristics are
established by random processes before birth, says John G. Daugman of the
Computer Laboratory at the University of Cambridge, who developed the algorithm
behind the process. The iris's pattern--which is different in each eye--appears
to persist virtually unchanged throughout life. Even identical twins have
unique iris morphology.  What is more, no prosthesis can defeat the system
because it detects the minute pulsations and pupil changes that indicate living
tissue, contends Sensar's Kevin B.  McQuade.

  Experts in high security have shown a keen interest in iris-based
identification: McQuade speaks in hushed tones about inquiries from the Central
Intelligence Agency. Frank Bouchier of the Security Systems and Technology
Center at Sandia National Laboratories tested an early version on 199 eyes and
found zero false accepts and less than 5 percent false rejects.

  The first ATMs equipped with iris recognition are expected by the end of this
year. And if the technology catches on, it could protect users of "smart"
cards. The customer's iris code could be stored on the card, and the merchant
would be unable to access the data unless the customer--or more precisely, the
customer's eye--were present.