But unless a biometrics technology vendor has that kind of database, it's difficult to do the kind of testing required to make sure their products really work.

To circumvent that problem, the University of Calgary recently opened a biometrics technologies lab where it will use "inverse biometrics" to conduct its research.

"Instead of doing analysis, we are doing synthesis," said Svetlana Yanushkevish, a University of Calgary professor and head of the Biometrics Technologies Lab. "We generate synthetic images, such as faces and fingerprints, collect them in a database and then we can consider them as live, statistical data to test biometric devices."

Synthesis is the process of creating mathematical models of real images. That approach helps solve the problems posed by privacy regulations, Yanushkevish explained.

"If some vendor of biometric devices says the performance of the device is 100 per cent, you have to ask how much real data they use to test their devices, because usually it's very restricted," she said. "You can't collect fingerprints for a live database and if you want to test it thoroughly and carefully you have to have a live database."

The lab, which recently bought a biometric workstation from Oakville, Ont.-based Comnetix Inc. to power the research, will also focus on synthetic aging of faces, research Yanushkevich says will benefit security personnel in banks and airports in particular.

"You can't imagine what people can do," said Yanushkevich, who is co-authoring the first book on the inverse problems of biometrics, due to be published next year. "They can do surgery to change their face, they can put artificial burns on them … people are very inventive when it comes to forgeries."

The Comnetix workstation, which is part of a contract valued at almost $100,000, will capture and process digital fingerprints. As well, the computer comes with Neven Vision's facial recognition technologies.

Steve Poelking, director of research, infrastructure and applications at IDC Canada, said security at national border points is a major issue for governments, Crown corporations, and other regulated entities. It's essential for governments to make sure they have the right IT systems -- as well as the proper policies and procedures -- in place in order to transmit, correlate and interpret disparate data and intelligence, he said in an e-mail interview with TIG.

"IT security labs serve a valuable function in passing and strengthening those IT security systems before going live in the real world," said Poelking.

According to Alan Brousseau, vice-president of investor relations and business development for Comnetix, the vendor typically sells its systems to law enforcement, although it also targets the health care, nuclear facilities, airlines, gaming and financial services sectors as well.

Although experts predicted that biometrics would be a routine part of life after the Sept. 11, 2001 terror attacks, that hasn't happened. That's because the industry is still very young and is comprised of many small players with a lot of competing technologies, said Brousseau.

"Governments have been hesitant to spend a lot of money without knowing their systems are going to be around for a while, and there have also been some privacy concerns in Canada."

People are hesitant to offer their fingerprints because they tend to associate biometric fingerprint technologies with the criminal justice system, he said, but there seems to be less concern about facial recognition biometrics.

"They're used to getting their pictures taken and their pictures are in databases already, so they're not as apprehensive about undergoing facial recognition tests," said Brousseau.

But facial and iris recognition technologies are not as mature or accurate as fingerprint biometrics at this point, he said.

"There is no huge iris database of criminals, but we have big fingerprint databases we've been developing for 100 years now."

But public sector agencies tasked with choosing biometric technologies and vendors have to know to ask the right questions, said Amanda Goltz, a consultant with the New York-based International Biometric Group.

"You have to ask how many false matches there are," she said. "How many times did it say this was a person when it's not; how many false rejections are there? How many times does it reject the true and accurate person?

"A question many people forget to ask, and that vendors may or may not volunteer, is how many people fail to be able to enrolled; how many people can't take part because their fingerprint can't be scanned?

That's an important issue, because it influences the accuracy statistic, she said.

"Maybe it works 100 per cent of the time on the three people you managed to put in the system."