News

The technology Canada is planning to use to guard against passport theft and fraud involves a series of confounding algorithms that will help confirm every traveller's identity – without disturbing the traveller.

Canada plans to spend $10-million on "facial biometrics" – the method of identification chosen by the International Civil Aviation Organization to harmonize passport information globally. The investment is part of a $700-million plan for fighting terrorism that Ottawa unveiled this week.

Biometrics uses advanced computerized recognition techniques to identify people by the facial features unique to them.

"Initially, we were looking at a way to speed up the flow of passengers at airports," ICAO spokesman Denis Chagnon told globeandmail.com. But the need to have machine-readable travel documents became more urgent after September 11, and the organization adopted a standard for facial biometrics last May.

Border authorities and airport staff have always relied on faces to confirm identity by simply eyeballing a match between a person and their passport photo.

The new technology will do the same thing but at a much deeper level. It involves three components: a camera or "reader" to capture the person's image, computer chips embedded in passports to carry information about the passport holder and a database to verify the passport's authenticity along with that of its bearer.

"Facial recognition is really the most unobtrusive use of the technology, Ashley Kelly, director of marketing at biometrics manufacturer Acsys Biometrics Corp., told globeandmail.com.

Although fingerprinting and iris recognition are precise, both require individuals to make contact with readers – something people may find intrusive.

Iris recognition can also pick up information, such as whether a woman is pregnant or if someone is using drugs, which is considered by some to be a violation of privacy.

Mr. Kelly says facial technology mimics the way the human brain identifies other people.

When we look at a person, "we don't know that there's precisely two inches between their eyes, we identify them in terms of differences with others."

The largest and most expensive part of implementing the system at borders and airports involves building databases to hold information about every passport holder.

A consistent programming framework will allow countries who implement the ICAO standards to share that information.

Biometric recognition uses "local feature analysis", a process in which the software locates the eyes, then maps out the entire face using various measurements between features.

The resulting series of measurements would be given a corresponding number that would be stored in the database as the person's I.D.

There is a weakness to basing the analysis on a single image, however: If a person presents their face at a different angle at check-in time, their measurements may no longer correspond to those in the database.

An alternative is to take multiple images and fold them all together into what is known as a template or "memory engram" and store that template into the database. The template could be used despite variables such as a haircut or glasses.

A computer chip will be embedded into each passport and include important identification information, such as name, birth date, passport number, as well as the facial template.

The chip would be encrypted to prevent it from being forged or altered.

These chips – and passports – would be readable by machines in every country that adopts the ICAO standard.

The least high-tech and most inexpensive component of the system is the camera or "reader" that quickly takes an image of the live passport holder.

But ultimately, the system preserves a human element. There will always be customs officer doing a crosscheck.

"Rather than identification, the focus here is on verification -- verifying one to one that you are who you say you are," Mr. Kelly said.