A broad range of news services are now covering a technology called facial recognition (FR). Maybe you have heard of it or possibly you were scanned by it but didn't know it. Basically, FR technology is one of several emerging automated methods lumped under the category called biometrics that identifies people or authenticates their identity through a part of the human body. Biometric identification determines who you are by using captured information from physical characteristics such as facial information, DNA, a fingerprint, hand geometry, an iris or retina, or a voice to search a database for a match. For example, it compares your fingerprint to a database of fingerprints or a video capture of your face with a database of faces.

Biometrics can be used for identification and verification purposes. Identification involves performing a one-to-one comparison between a live scan and a stored template. The individual is then accepted or rejected depending on the percentage match score. Examples include identification of individuals for passports, identity cards, driver's licenses, computer security and access control systems. Verification uses many successive one-to-one comparisons against templates in a database. The hit list of potential matches are then returned, ranked by decreasing percentage. Scanning a crowd with a camera, using FR software and comparing the images to a database is an example of verification.

Facial recognition techniques are used successfully and without public outcry in several situations including entrance to buildings, or security for department stores, gambling casinos, and boarding areas in airports. Gambling casinos have used it for years to rapidly identify undesirable patrons, and are always improving their techniques. Recently, a gaming consulting company and a biometric solutions developer announced a partnership to combine ID tracking software and a database of undesirable patrons for all of North America, thus making the combined product more desirable to casinos that previously only had a locally captured database of patrons.

The police also have used the FR software with other programs to aid in the identification and capture of offenders. For example, the Porterville Police in Los Angeles county, California purchased a complete law enforcement suite that provides software with a criminal database and a facial recognition package that includes a computer-aided dispatch system, records, and mobile use version for police cars, to aid in the quick identification of criminals.

There are other possibilities for facial recognition technology such as physical access control, point-of-sale verification, PC access, eCommerce and Internet verification, as well as identity fraud prevention. These are seemingly innocuous applications and probably will be embraced by the general public.

Physical access control uses your face instead of a magnetic card or key for entry into your house or apartment. Some details will have to be worked out like what do you do if you want to let your friend into your house to check on the animals or plants? Presently, you can loan her your key but what can you do if you have a facial recognition security system?

FR for point-of-sale could be used for check cashing or credit card checks. For PC access control it could help eliminate the frustration of having to remember all those user names and passwords we need remember but write down and leave nearby, defeating the purpose of the intended security. eCommerce and Internet is another potential use for FR technology. You could log onto a site with only your face. This could help users move beyond their fear of purchasing items with their credit cards because of potential theft of their card number. According to facial recognition developers, ID fraud could be prevented with this FR technology.

But recently, a couple events led to more public scrutiny of this technology when crowds of people were looked at without their knowledge and compared to a list of known criminals. The first was at the Super Bowl and the other was in a renovated down-town district with restaurants and night clubs.

In January 2001, during Super Bowl XXXV in Tampa, Florida, a facial recognition system was used with 20 surveillance cameras to secretly scan and capture the faces of fans. Several algorithms were used to measure 80 facial features from the 100,000 images acquired, such as the distances and angles between geometric points on the face, which include the mouth extremities, nostrils, and eye corners. These measurements were then used to produce a face map. This map was then compared against a database of suspected terrorists and known criminals. For some FR packages, it is considered a match if 14 of 80 distinctive points align.

Graphco Technologies Inc., said the event in Tampa gave the company a chance to learn how the software would perform, which camera angles were most effective and how the lenses of the 20 video cameras should be focused in a public place. As each person passed through the four main stadium gates, a camera captured dozens of images, which were fed into computers. The computers compared the portraits against a database assembled from law enforcement agency files by a Massachusetts company, Viisage Technology Inc., which markets the software. The digitized images were constructed using 128 facial characteristics -- everything from the width of a nose to the angle of a cheekbone. Each apparent match at the game was designed to be flashed side by side onto a computer screen at a stadium command post. A police officer determined whether the faces were those of the same person. The turnstile images were then discarded.

In her exclusive article to ChipCenter, "Living on the Wrong Side of a One-Way Mirror: Face Recognition Technology and Video Surveillance," lawyer Grayson Barber says, "Cast strictly in terms of constitutional law, the legal case against video surveillance and face recognition technology rests on the Fourth and Fifth Amendments, which protect against unreasonable searches and seizures and self-incrimination. These constitutional protections have never been invoked against blanket surveillance of whole populations. Face recognition and video surveillance permit organs of the government to probe and search bad actors and good actors alike, without warrants and without probable cause or individualized suspicion."

The use of a biometric facial recognition system such as that employed at the Super Bowl is different in certain respects from other more familiar uses. FR technology makes human recognition a more automated, computerized process. It is this aspect of the use of biometrics that raises the fear that we are losing our ability to control information about ourselves and that we are losing our right to privacy. According to lawyer John D. Woodward Jr:

Does the use of this technology violate legally protected privacy rights? Legal rights to privacy may be found in three sources: federal and state constitutions (if the entity invading your rights is a government actor), the common law of torts (if the entity invading your rights is a private actor), and statutory law. Although the word "privacy" does not appear in the U.S. Constitution, the concern with protecting citizens against government intrusions into their private sphere is reflected in many of its provisions. For example, the Fourth Amendment protects against unreasonable searches and seizures. The constitutional "right to privacy" therefore reflects concerns not only for one's physical privacy - the idea that government agents cannot barge into one's home - but also for less tangible interests - the idea that citizens should be able to control certain information about themselves and to make certain decisions free of government compulsion. And the Supreme Court has cautioned that it is "not unaware of the threat to privacy implicit in the accumulation of vast amounts of personal information in computerized data banks or other massive government files."

The use of biometric facial recognition potentially implicates both types of privacy interests. Nevertheless, law enforcement's use of the technique at the Super Bowl does not appear to run afoul of the protections afforded by the U.S. Constitution. Some civil libertarians argue that the sort of mass, dragnet scanning that took place at the Super Bowl is improper, and that law enforcement must have individualized, reasonable suspicion that criminal activity is afoot before it can "search" a subject's face to see if it matches that of a wanted individual in its database.

Under current law, however, the type of facial recognition used at the Super Bowl would almost certainly be constitutional. The Supreme Court has explained that government action constitutes a search when it invades a person's reasonable expectation of privacy. But the Court has also found that a person does not have a reasonable expectation of privacy with regard to physical characteristics that are constantly exposed to the public, such as one's facial features, voice, and handwriting.

So although the Fourth Amendment requires that a search conducted by government actors be "reasonable," which generally means that there must be some degree of suspicion that the person to be searched is engaged in wrongdoing, the scan of spectators' facial characteristics at the Super Bowl did not constitute a search. And with respect to concerns about information privacy, if law enforcement officials limited their actions to simply comparing scanned images of people entering the stadium with their computer database of suspected terrorists and known criminals, then information privacy concerns would probably not arise so long as no information about individuals were retained, disclosed, or linked to any other database.¹

Should You Be Concerned?

FR technology is in its infancy and it will certainly develop dramatically. Already, the gambling casinos are showing us what is coming as they join together and link databases of local casinos to others in North America, and eventually to casinos around the world. You might ask, "global linking of databases to protect their businesses, who can complain about that?" It all depends on how the information is used and possibly abused.

As facial recognition or other biometric databases become interconnected, the threat to information privacy has the potential to increase significantly. FR technology can lead to the loss of information privacy because your movements can be tracked (similar to tracking suspects by credit/gas cards) and your image captured clandestinely. Authorities (supposedly only those with legal access) could monitor your actions in real time or over a period of time. In an extreme example, authorities could place you in a database and use surveillance cameras to capture your facial image and digitally transmit this biometric information for instantaneous searching against a watch list. You could be followed as you go about your mundane, everyday tasks (that don't warrant watching) such as going to work, entering your office building, getting money from the ATM, making a purchase in a store, and a myriad of other tasks. If there are cameras, your image will be matched with information in the database, allowing someone to track your movements. Additionally, these cameras will catch not only you but anyone else that is with you, and then compare them to an existing database (that may be linked to many other databases). The interconnecting of databases and the use of cameras at the store, the ATM, the gas station, etc., could lead to an abuse of privacy rights. However, constitutional protections are available only against actions by government agents. For unwarranted surveillance technology used by a private entity such as a market research firm, you would have to seek legal recourse under common law privacy rights. If the surveillance only tracked the location and activities of individuals in public, it probably would not violate your common law privacy rights. On the other hand, Congress could intervene and pass legislation to regulate such practices.

Recently, House Majority Leader, Dick Armey, R-TX and The American Civil Liberties Union made the first step recently by releasing a joint statement about their concerns on the proliferation of surveillance devices. This is an interesting partnering because Majority Leader Armey is considered an ultra conservative and the ACLU is viewed as very liberal. They make mention in the statement the events surrounding what some call the "snooperbowl" in Tampa, and of the plans by the Department of Motor Vehicles in Colorado to create a database containing 3-D images of applicants for driver's licenses. They are concerned that Colorado could misuse the information to identify, track, and store the movement of its citizens.

Considering these potential concerns, we need to guard against developing technology without controls, developing something just because we can. It is appropriate that we watch the watchers and we be ever vigilant of emerging technologies that may invade our privacy, so we can gather our wits and prevent potential abuses. At the same time, we should also realize that biometric technology has many positive applications, we just need to watch it not squash it.

Reference:
1. Super Bowl Surveillance: Facing Up to Biometrics; John D. Woodward Jr.

More information:
1. The Biometrics Consortium.

2. Links to Biometric Sites.

3. Police video cameras taped football fans.

Vote in the Poll

Wednesday July 04 02:31 PM EDT
Tampa Scans the Faces in Its Crowds for Criminals
By DANA CANEDY The New York Times

Police in Tampa, Fla., are using cameras equipped with face-recognition software to search for criminal suspects among people in a downtown district.

TAMPA, Fla., July 3 Camera shy? Then steer clear of Tampa's nightlife district. The Tampa Police Department has placed three dozen security cameras with face-recognition software in a downtown district popular with locals and tourists. Now, everyone who visits the district, Ybor City, for a burger or a beer runs the risk of having his face digitally scanned and the noses, cheeks and chins checked against a mug-shot database of murderers, drug dealers and other criminal suspects with arrest warrants.

The police have used surveillance cameras in other cities to record and catch criminals in the act. But Tampa's effort is the widest by a police department in this country to fish for criminal suspects in a general-public sea, using this technology.

The makers of the system, the Visionics Corporation of Jersey City, offered Tampa free use of it for a year, in an effort to build a market among municipalities. City officials, who had used a competing system in January to scan the crowds at the Super Bowl for possible terrorists, were agreeable.

"It's a public safety tool, no different than having a cop walking around with a mug shot," said City Councilman Robert F. Buckhorn Jr., chairman of Tampa's public safety committee.

Besides, he added, on a Ybor City street of restaurants, nightclubs and stores crowded with 20,000 people, "your expectation of privacy is somewhat diminished, anyway."

Randall Marshall, legal director for the American Civil Liberties Union of Florida, disagreed, saying it amounted to subjecting the public to a digital lineup.

"This is yet another example of technology outpacing the protection of people's civil liberties," Mr. Marshall said. "It has a very Big Brother feel to it."

The system, in effect since Friday, has not yet identified any suspects, though the system used at the Super Bowl identified 19 people thought to be wanted on outstanding warrants for misdemeanors.

Reaction in Ybor City was mixed.

Jason Skinner, a security guard buying sandwiches at a deli across the street from a camera mounted to a utility pole, said that, despite his occupation, he opposed the digital peeping by the police.

"It's invading people's privacy," Mr. Skinner said of the camera aimed in his direction. "They're all over the place."

But some Ybor business owners said they hoped the cameras became a permanent fixture, much like street lights.

"I don't find it an invasion of my privacy, and my customers don't either," said Jill Wax, 48, the longtime owner of La France vintage clothing store and the former president of the Ybor City Chamber of Commerce. "My only problem with it is how many people will they catch, how beneficial will it be? I guess we'll have to wait and see about that."

Since the system, called FaceIt, started, police officers in a nondescript command center in a neighborhood building, monitored a bank of television screens filled with faces in the crowd, zooming in on individuals and programming the equipment to scan them.

The FaceIt computer broke down each facial image into something similar to a map, with 80 reference points to check. If the system matched more than a dozen of those points against an image in its database which is to include 30,000 faces when it is fully operational next month it would signal a match.

At that point, a system operator would determine if the images were similar enough to radio a uniformed officer, who would investigate and possibly make an arrest.

No suspects were identified today, though several people were caught licking ice cream, gulping sodas and stuffing their mouths on their lunch break.

Detective Bill Todd, in charge of the operation, said it was not as if people were on "Candid Camera." The department has placed signs in the area warning passersby that "Smart CCTV is in use," referring to closed circuit television, but signs are not visible from every area in which the cameras operate, and most people interviewed did not know what the message meant.

Similar technology is used by banks, casinos and other businesses. But Samir Nanavati, a partner at the International Biometric Group consulting firm in New York, which advises companies that consider using such technology, and others said that people generally expect some form of surveillance in those settings and can decide ahead of time whether they want to be on camera.

"The question is, can they educate people in that area sufficiently enough so that they understand what is taking place?" Mr. Nanavati said. "And, if they do understand, is there any way for them to opt out and choose not to utilize the technology and the answer is no."

The Tampa police call the privacy issue overblown because the camera does not record images of people who have not been charged with a crime. "We are not cataloguing a thing," Detective Todd said.

"If the image the camera takes is not in the database," he said, "it immediately dumps the image."

Both the manufacturer and the police say the chance of a false arrest based on the facial scanning is slim and an acceptable trade-off for the possibility of nabbing a criminal who might otherwise remain at large.

"We expect the police to see results," said Joseph Atick, president and chief executive of Visionics. "For criminals who object to it or have a warrant out there will be a deterrent factor from people saying `I don't want to mess with this.'"

Detective Todd said that the expenses of operating the system, which would cost about $30,000, were nominal, because the department would use officers who would otherwise patrol the streets.

He said the system let the department "maximize the process of pointing out people we're looking for without putting 20 more officers on the street and looking for those same people."

FaceIt has been used since 1998 in the London Borough of Newham, whose officials have attributed a drop in crime to it. Last month, it was introduced at Keflavik Airport in Iceland.

But Atlanta and other cities have rejected the system because they were not convinced of its effectiveness or accuracy.

Tampa first used face-recognition surveillance in a competing system at the Super Bowl at Raymond James Stadium in January. Even though the system spotted 19 people thought to be subjects of outstanding warrants for minor crimes, none were arrested because the crowd was so large and because the number of matches exceeded the police's expectations.

"We thought we were ready to use it, but getting through the crowd and the architecture of the stadium proved overwhelming," Detective Todd said.

The detective said Ybor City had a crime rate about that of the city over all. But the area, he said, had a higher percentage of "crimes of opportunity" muggings, purse snatchings and the like in which criminals focus on areas drawing large crowds, making it a good neighborhood to test the system.

Gil Rizzo, 42, an account representative from Tampa relaxing on Seventh Avenue, Ybor's main drag directly in front of one of the cameras agreed.

"I'm in favor of it because of the security," he said. "A lot of nights, there has been shoplifting, women got mugged and robbed. It's safer because of the cameras."

Vote in the Poll

Analog Main | Product of the Week | Columns | Editorial | Tech Notes

		Click here to get your listing up.