ID System Gets in Face of Criminals

The Los Angeles Police Department is seeking half a million dollars from the federal government to expand the use of advanced facial-recognition systems to identify criminal suspects.

Police officials say the technology could be an important step in fighting crime.

“It’s like a mobile electronic mug book,” said Rampart Division Capt. Charles Beck. “It’s not a silver bullet, but we wouldn’t use it unless it helped us make arrests.”

Civil liberties advocates are less enthusiastic about the technology, questioning its reliability and the privacy issues it raises.

The potential of the facial-recognition technology could be seen in a recent police stop on Alvarado Street just west of downtown Los Angeles, where police have been testing the cameras.

Two young men were illegally riding double on a bicycle. LAPD Officers Mark Hubert and David Nick suspected the two were gang members. If so, they might have been in violation of an injunction barring those named in court documents from a variety of activities in the area, including gathering in public.

As they questioned the pair, Rampart Division Senior Lead Officer Mike Wang pointed a hand-held computer with a camera attached toward the man on the bicycle seat.

Facial-recognition software in the device compared the image with those in a database that includes photos of recent fugitives, as well as 78 members of the Mara Salvatrucha gang and 45 members of the 18th Street gang. The groups are two of the city’s largest Latino gangs, both with bases in the area covered by the LAPD’s Rampart Division.

Within seconds, the screen had displayed a gallery of nine faces with contours similar to the man’s. The computer concluded that one of those images -- of Jose Hernandez, an 18th Street member subject to the civil injunction -- was the closest match, with a 94% probability of accuracy.

It was enough to trigger a search, and officers say they found a small amount of methamphetamine on one of the men -- later confirmed to be Hernandez.

Hernandez, 18, was arrested on suspicion of violating the injunction by possessing illegal drugs. The city attorney’s office says a decision about whether to charge him is pending. Attempts to reach Hernandez for comment were unsuccessful.

Hernandez is among 19 suspects arrested so far as a result of the Rampart Division’s use of the cameras, which were donated and are still considered experimental.

The LAPD has been using two of the computers donated by their developer, Santa Monica-based Neven Vision. The firm, a pioneer in facial-recognition technology, was looking to have its products field-tested.

Among those leery about police use of the devices is Ramona Ripston, executive director of the American Civil Liberties Union of Southern California. “This is creeping Big Brotherism.... There is a long history of government misusing information it gathers,” she said.

Ripston voiced concern that the cameras would encourage racial profiling and might lead officers to question any youngsters who dressed in certain ways.

In the eight weeks that the Rampart Division has been testing the devices, they have been used about 25 times, police said, resulting in 16 arrests for alleged criminal contempt of a permanent gang injunction and three on outstanding felony warrants.

On one occasion, Wang said, the technology cleared a man whom officers suspected of being someone else.

So far, the city attorney has filed seven injunction cases in connection with arrests assisted by the technology. A judge dismissed one case after questioning the technology, but it was refiled. Two of the cases have been resolved with guilty pleas.

Luis Li, chief of the Los Angeles city attorney’s criminal branch, said the technology had presented few problems in court, because it was used only as an initial means of identification. Before someone is charged, authorities run fingerprints that become the basis for confirming a suspect’s identity.

Li said his office had no privacy concerns over the technology. “If you are standing in the street, you have no expectation of privacy,” he said.

Ripston also questions whether the results are accurate.

“I understand there are very real gang problems in parts of Los Angeles that people are anxious to address,” Ripston said. “But this technology is unproven. There is a great possibility of mistakes, and cameras are operated by human beings with all their own prejudices.”

Other experiments with facial-recognition software have yielded mixed results. City officials in Tampa, Fla., pulled the plug on such a system last year in the Ybor City entertainment district because it failed to yield arrests.

At Boston’s Logan International Airport in 2002, two separate systems failed 96 times to identify volunteer “terrorists.” The systems correctly identified 153 other volunteers.

But on Florida’s Gulf Coast, the Pinellas County Sheriff’s Department used $8 million in federal funds for facial-recognition technology to be placed in jails, an airport and squad cars. Since June 2002, inmates have had their facial images preserved digitally.

“Almost daily, a John Doe or someone claiming to be someone else is identified,” said Lt. Jim Main, who oversees the Pinellas project.

Ripston said research had also raised concerns about how suspects’ aging might affect the systems. A National Institute of Justice study found that digital comparisons of photos taken 18 months apart led to false rejections 43% of the time, she said.

Experts say the technology works best in good light, with a static subject at close range.

Hartmut Neven, developer of the software the LAPD is trying out, says his system uses an algorithm to translate various parts of the face into complex mathematical patterns employed to develop unique numerical templates. His technology concentrates on the areas beneath the hairline and above the chin, so beards and different hairstyles don’t influence the outcome, he said.

When a new image is taken, it is compared to a database of images for similarity and differences, and those that best match are generated as a gallery in order of likely accuracy.

Neven said movement, poor lighting and distance from the subject could make recognizing a face harder. But, he said, it is the cameras and their lack of resolution that present the biggest obstacle.

“As the image quality improves, you can do more,” he said. “We’ll be able to do the texture of the skin, scars and other marks, and eventually irises.”

Still, the technology remains in its infancy.

“At present, no one has accurately used a facial-recognition system to pick faces out of a crowd,” said Jim Wayman, director of San Jose State’s Biometric Test Center. “But as an investigative tool, like in the case of the LAPD, it can be useful.”

The Rampart Division has been among the most eager in the city to employ new technology. In addition to the facial-recognition cameras, the division has used surveillance cameras donated by General Electric Co. to monitor the MacArthur Park area. It was GE’s scientists who later introduced officers to software developer Neven.

The LAPD has made some attempt to combine the two technologies, using facial-recognition software to try to identify people from images recorded by the MacArthur Park surveillance cameras.

Wang recently demonstrated the procedure. As he used a joystick to zoom in on a man captured on video at a bus stop near the park, the software began a comparison. As the man moved, however, the image quality deteriorated to a point that it didn’t allow a comparison. But Wang remains hopeful.

“It is only a matter of time,” he said. “I believe sooner rather than later, we’ll make it work.”