An overlay depicts how facial recognition software pinpoints certain features and compares them to a database. The distance between the eyes and the shape of the cheekbone can be used to determine characteristics like sex and age.The Globe and Mail/The Globe and Mail
A decade-old effort by Canadian officials to use facial-recognition software to analyze passport photos is credited with helping catch a man accused of providing bogus travel documents to criminals.
Last Friday, the RCMP National Division arrested 47-year-old Harbi Mohamoud Gabad of Gatineau, Que.
According to the indictment filed at the Ottawa courthouse, Mr. Gabad is facing multiple criminal counts, including making fraudulent passport applications for organized-crime suspects from Quebec and British Columbia's Lower Mainland.
The charge sheet linked Mr. Gabad to bogus travel documents used by Rabih Alkhalil.
Mr. Alkhalil, who was arrested in Greece in January 2013, is wanted in Canada in connection with two high-profile 2012 homicides: the fatal shooting at an ice cream parlour in Toronto's Little Italy of John Raposo and the slaying of Sandip Duhre at a restaurant in Vancouver's Sheraton Wall Centre.
Mr. Gabad is also alleged to have helped one of Mr. Alkhalil's brother, Nabil, who has a record for cocaine trafficking and assault with weapons.
Another crime figure Mr. Gabad is accused of helping procure a passport is Shane Maloney, who was arrested in 2012 on allegations that he was part of a syndicate that united Quebec and B.C. criminals to import large amounts of cocaine to Quebec and Ontario.
According to a search warrant application for Mr. Gabad's residence obtained by the CBC, the investigation involved facial recognition technology that detected in 2012 that there were two passports issued to Mr. Alkhalil, one in his name and another under a bogus identity.
Facial recognition is part of the security measures initiated by the passport office following the Sept. 11, 2001, terrorist attacks.
"A combination of trained passport officers and specialized technology is used to verify applicant identity through data authentication and facial recognition," Citizenship and Immigration spokeswoman Sonia Lesage said in an e-mail reply to the Globe.
She said Passport Canada has been using the technology to detect fraud since 2009
However, the project has been in development for at least a decade.
The Office of the Privacy Commissioner of Canada (OPC) says that it has been asked to review the impact of Passport Canada's use of facial recognition since 2004.
By 2006, Passport Canada was looking for a computer contractor to device a facial recognition system that could handle millions of photos, according to a tender posted on the MERX website.
Initially, the system would deal with one million scanned passport photos in its database, with about four million images being added annually during the next five years, the tender specified.
The software would have to be available 24 hours a day for about 100 primary users within Passport Canada's Security, Policy and Entitlement Division.
Photos would be analyzed for two types of verifications – "1:1" checks, where an applicant's photo is compared with an image already in the database, to make sure the applicant is really the person she or he claims to be; and "1:N" checks, where a photo would be compared with all the other pictures in the database to detect whether the applicant is already registered under another name.
Typically, a facial-recognition software measures and compares features such as the distance between the eyes, the width of the nose or the length of the jawline.
Such analysis is assisted by the new guidelines of the International Civil Aviation Organization, which specify that machine-readable travel documents have to use photos with neutral facial expressions.
According to an OPC review in 2006, Passport Canada test runs showed that the software made a correct match in 88 per cent of the time if presented with good-quality photos that complied with the no-smile rule. If lower quality shots were used, the results dropped to 75 per cent.
"When an anomaly is detected, the final determination of whether the photo merits further investigation is made by a specially trained operator after further examination," Ms. Lesage said.
As facial recognition technology is further developed, opportunities to incorporate it into federal surveillance programs could increase, the OPC said in a report last year.
"For example facial recognition technology could be added to existing video surveillance systems, such as those used by the RCMP on Parliament Hill, the Canadian Air Transport Security Authority in airports, and by the Canadian Border Services Agency at all border crossings."
Tu Thanh Ha