Corrections or additions?
This article by Barbara Fox was prepared for the February 25, 2004 issue of U.S. 1 Newspaper. All rights reserved.
For Secure ID, the Eyes Have It
`Angels & Demons," the thriller that preceded Dan Brown’s "The Da Vinci Code," starts out with the murder of a scientist working on hotter-than-nuclear material. The thief carves out the man’s eyeball and uses it to gain entrance to a laboratory guarded by a retinal identification device.
That could never happen, scoffs David Johnston of LG Electronics USA. That’s because both retinal identification technology and the technology that he sells, iris identification devices, can easily discern whether the eye is alive.
It takes just two seconds for an iris recognition device to riffle through millions of datasets to identify the digital representation of one person’s iris, the circle around the pupil of the eye. Like snowflakes, no two irises are alike.
Korea-based LG Electronics was the first company to license and produce a commercially viable iris recognition platform, and it has second generation products in more than 1,000 locations on six continents. LG licensed this technology from Moorestown-based Iridian Technologies, the company that resulted from the stock-swap merger of IriScan and Sensar, a Sarnoff spinoff.
On Cranbury-South River Road, David Johnston heads worldwide marketing for LG’s iris recognition technology division, and in one year the office and warehouse space has doubled in size, to 15,000 square feet. Of the 45 employees in the division, about 15 are working in Cranbury at any one time.
Increased attention on security is revving up the market for all the biometric methods, including fingerprints, voice prints, and facial or hand identification. Johnston believes the technology of iris recognition can do more than keep us safe, that it has the potential to change the way we live.
"I happen to believe that human authentication will be as commonplace as using the web is today," says Johnston. "Most people will start accessing the Internet from wireless devices, and it will be very important to know who is accessing what."
Early investors in the technology included Penny Lane Partners, the SBIC venture capital fund on Palmer Square that put $1.75 million into the company now known as Iridian Technologies (www.iridiantech.com). "Compared to all the other biometrics out there, iris recognition is superior in all ways," says Penny Lane’s Steve Shaffer. Justifiably disappointed that Penny Lane’s 1998 investment has not yet paid off, Shaffer hopes Iridian will be sold soon and says it is being marketed by a New York investment firm, Broadview International. "We are a little bit upset that it didn’t gain traction any faster than it did," says Shaffer.
A major drawback to current iris recognition technology is that cameras are expensive, because not enough of them are being built. That may soon change, partly because Iridian is busily licensing its technology to companies like LG, Oki, and Panasonic. And partly because the original patents on using the iris as a biometric identifier will expire in the next couple of years. "When the patent expires, the field could explode, and then other companies can use the iris but write their own algorithms," Shaffer predicts.
Here’s how it all started. In 1987 ophthalmologists Leonard Flom and Aran Sofir won a patent on the original idea, to use irises for identification. In 1991 a Harvard-trained researcher at Cambridge University, John Daugman, wrote the algorithms for converting the image of the iris into a compact digital code. In 1995 IriScan fulfilled a contract to do a prototype unit for the U.S. military, and both IriScan and Oki Electric Industry released commercial products.
To this technology Sarnoff contributed a way of looking at the eye. "Among the technologies we developed for Sensar were methods for acquiring any iris image – finding the iris and getting an image," says a Sarnoff spokesperson. Daugman converted that to a biometric template (his images of irises are on the cover of this issue).
By 1997 Sensar was raising money, mostly from banks, to develop a product for Automated Teller Machines. Sensar was licensing the Flom/Sofir patent and the Daugman algorithms from IriScan.
Also that year LG Electronics and IriScan signed a co-development and distribution agreement. Founded in Seoul in 1958 and known in the United States as the company that bought Zenith, LG has four business groups: consumer products, electronics, chemical, and financial. LG Electronics USA employs 250 people at its North American headquarters in Englewood Cliffs, and Charlels Koo is the chairman and CEO of Johnston’s division, which deployed its Iris Access platform in 1999.
When IriScan bought Sensar in 2000, the company name changed to Iridian Technologies, and it attracted an additional $33.5 million in funding. Licenses to Matsushita, Panasonic, and Lockheed Martin followed. Now the technology has been deployed not only at two dozen airports around the world, but also at such smaller sites as the Plumsted schools in New Egypt (a pilot project to restrict access to school buildings) and a hospital in Eagleville, Pennsylvania (to prepare for the data privacy and computer security requirements of the Health Insurance Portability and Accountability Act).
Clean labs for pharmaceutical firms are among LG’s biggest customers, and that’s partly due to the cost. "At what we cost, we can’t afford to put us on just any door," says Johnston. For a medium-sized enterprise, to have a unit on four doors costs $16,000 plus integration costs and installation. For more doors, the cost per door drops.
Visitor or contractor identification at data centeris another important use. "Virtually any company understands the value proposition we afford," says Johnston. "Many compromises of corporate information occur from individuals that are within the organization."
Banks were initially interested, and indeed much of Sensar’s funding came from its financial institution investors. Oki employs the technology for banks in Japan. But in the long run, says Johnston, U.S. banks have been unwilling to pour money into ATM machines because of the extra equipment and bandwidth that would be needed.
Border crossing applications are where the real money is, says Shaffer of Penny Lane Partners. "The United States is not an early adaptor of this systems, and I don’t know why. It is working in Frankfurt, Amsterdam, Afghanistan, United Arab Emirates, and Canada. The hardest part of the system is making sure the person is who he is when you enroll him."
Here’s how it works: Using infrared light, the camera focuses on your iris’s distinct features – rings, dark areas, spots, striations and so on – even though you may be wearing glasses or contact lenses. The algorithms convert the camera’s image to compact digital codes. The code can be compared with any other iris code by means of a mathematical technique known as the Hamming distance. This extraordinarily fast calculation can do millions of comparisons per second on easily available hardware. The difference between any two codes which come from different individuals will be very large, whereas fingerprint analysis can be more ambiguous.
Iris ID is working now. At the JFK airport on Long Island terminal employees get access to the tarmac by looking into a camera. At a laboratory in Delaware, scientists working on top secret formulas are gowned, gloved, goggled, and masked. They gain entrance to the lab without touching anything, merely by looking at the door. At a medical center in Boston, employees enter the disposal center for radioactive medical refuse just by looking at a camera. It can also be cost effective for correctional facilities, which use only a limited number of entrances.
Another option that is favored by countries in the European Union, where privacy is the clarion call, is to eliminate databases with iris codes and embed the code into a passport or a smart card. Insert the card and the camera need only decide whether you are the person who is supposed to have that card. The individual has control over his biometric profile.
It’s accurate because, unlike your retinas, your irises don’t change after you are one year old. In a test done in the United Kingdom, Daugman’s algorithms were compared to more than two million samples, and they produced no false matches. No two irises have been found to be identical, but the "one in a million claims" have receded.
"Iridian used to make claims that its iris recognition technology had an ‘equal error rate’ (the point at which the false-match rate equals the false non-match rate) of 1 in 1.2 million," says Trevor W. Prout, director of marketing for the Manhattan-based International Biometric Group (www.biometricgroup.com). "Iridian has since stopped making such theoretical claims in its marketing, which we believe were counter-productive because they set unrealistic expectations of how the technology may perform in the real world. That said, iris recognition is indeed highly resistant to false matching. However, it is not impervious. Also, more testing needs to be done to determine how the technology will perform as the size of the database grows very large."
There can be, nevertheless, problems with logistics. An airport in Charlotte had to install a special one-person revolving door to prevent interlopers from ducking behind someone who had passed the test. And if the smart-card iris technology does fail, the unwary traveler may not be carrying alternate identification.
Then there is the privacy question. Naysayers may cry "Big Brother!" and "Privacy invasion!" but that’s where iris recognition shines, insists Johnson. Whereas surveillance cameras can sweep a crowd to look for faces, iris ID only authenticates. "Every situation we have is an ‘opt-in’ one, to get through a door to do a job, to get fast pass, to get access to medical records. I don’t see how you have a privacy issue if it’s opt-in," he says.
Iris recognition can, in fact, combat identity theft, the fastest growing crime in America. "This is not Big Brother watching you, this is your last chance to look out for yourself," says Johnston.
Biometric technology in general has a long way to go. Another Princeton firm, VeriVoice, went under before making a commercial success of voice recognition technology. And with 52 percent of the business, fingerprinting is still the most popular biometric identification method, says the International Biometric Group. Facial and hand recognition have 11 and 10 percent respectively, and voice recognition has four percent, compared to iris ID with just 7.3 percent. The worldwide market for iris ID last year was $36 million, this year will be $72 million, and by 2007 is projected to be $283 million.
LG is working to capture the worldwide market, but it has plenty of competitors both globally and in the U.S., and the Flom/Sofir patents are due to expire. "It will be exciting times in this business," says LG’s Johnston. He professes not to be worried, because the Daugman patents will not expire for seven more years. "There is value in the proven success of the Daugman algorithms," says Johnston, pointing to companies in Asia and Eastern Europe that have come up with their own formulas and have achieved less than accurate results. "The beauty of the Daugman algorithm is that it has been rigorously tested and is highly accurate."
Johnston is convinced that sales will follow on the heels of need. He tells of the crowded border crossing between Malaysia and Singapore. Malaysians register with the Singapore authorities, pay the toll, look into the machine, "and in 30 seconds they do what used to take six minutes."
"That kind of passenger processing is really important," says Johnston. "Now 1.2 billion people fly in airports, and that will go to almost three billion by 2012. We will have three choices: build more airports, require passengers to arrive five hours before check in, or figure out a way to help process passengers."
Johnston spent most of his career in the advertising business before moving into high tech. A graduate of McMurray College in Illinois, Class of 1972, he joined the U.S. Merchant Marines and worked for a defense contractor before going to New York to work at J. Walter Thompson. As part of the WWP group, he did tours in the Far East (in Singapore, Malaysia, and Japan) followed by a stint in Paris, working on the global haircare business for Unilever. A single parent, he lives in Swarthmore and co-parents three school-age daughters.
"Serendipity led me to meet the CEO of Iriscan, and I was fascinated by the ubiquitous potential of the technology," says Johnston. "Why I love working for LG: We have an ‘intrapreneurial’ group inside a very large company, working on a product with global scope, and I am responsible for this product line worldwide."
He has no regrets about leaving the more glamorous world of advertising. "In advertising, you never knew when you would get another big account. I miss the diversity of people in ad agencies – they’re the quirkiest people in the world – but winning business is pretty exciting no matter what category you win it in."
– Barbara Fox
Corrections or additions?
This page is published by PrincetonInfo.com
— the web site for U.S. 1 Newspaper in Princeton, New Jersey.