Articles on Biometrics

And Now, the Good Side of Facial Profiling

John D. Woodward Jr.

This opinion article appeared in the Washington Post on February 4, 2001.

It seemed like an Orwellian revelation: Last week, law enforcement officials at Super Bowl XXXV in Tampa secretly scanned spectators' faces with surveillance cameras and instantly compared their "faceprints" against those of suspected terrorists and known criminals in a computerized database. Alarmed civil libertarians quickly raised the specter of a Big Brother government spying on its citizens.

But is the growing use of this technology cause for alarm? Is it an undesirable invasion of individual privacy, or does it represent a positive advance in security measures that generates benefits for society? As someone who closely follows law and policy issues related to biometrics -- technologies that use a person's physical characteristics or personal traits for recognition -- I believe we must not move precipitously to condemn a technology that can serve as a useful tool in the fight against crime and terrorism.

The technology used at the Super Bowl is known as facial recognition. Making an identification by looking at a person's face is a standard technique: Police regularly use mug shots to identify criminals, and most of us rarely go through a week without having someone ask, "May I see a photo ID, please?" But biometric facial recognition, which uses measurable facial features, such as the distances and angles between geometric points on the face -- the ends of the mouth, the nostrils and eye corners -- to recognize a specific individual, is a highly automated, computerized process. And as such, it raises real fears that we are losing the ability to control information about ourselves -- that we are being robbed of our anonymity and our privacy.

These fears are spawned by two aspects of biometric facial recognition -- clandestine capture, which means that facial recognition systems can scan a person's face surreptitiously, without their permission; and tracking, which refers to the fact that the technology makes it possible to monitor an individual's actions over a period of time.

At its most extreme, tracking could become a kind of "super surveillance" that allows the tracker both to "follow" a person in the present and to search databases to learn where he was months ago. For example, suppose the authorities placed me in their "watch list" database as someone they wanted to keep an eye out for. Surveillance cameras capturing my faceprint as I go about my many daily tasks would digitally transmit this biometric information for instantaneous comparison with the watch list.

As I board the Metro on my way to work, enter and exit my office building, stop by the ATM, or attend a political rally, a match will be made, allowing the tracker to know my movements. Similarly, the authorities can enter on their watch list the biometric information -- the faceprint -- of all those who attended the political rally with me and conduct searches to try to identify them from their movements. If such a system were established, it would become possible to compile a comprehensive profile of an individual's movements and activities. And the information from such tracking could be combined with other personal data, acquired by other means (like using someone's Social Security number), to provide even more insight into a person's private life.

But while all these fears are understandable, we should not allow perceived or potential threats to our privacy to blind us to the positive uses of biometric technologies such as facial recognition. Perhaps Osama bin Laden's henchmen were nowhere to be found in Tampa's Raymond James Stadium, but law enforcement officials at the Super Bowl were taking prudent steps to identify them if they were.

The national security community believes that facial recognition can also help it in identifying and protecting against threats to U.S. forces and embassies abroad. If a known terrorist can be identified before he closes in on his target, lives can be saved. In the wake of the terrorist attack on Khobar Towers, the Pentagon's Defense Advanced Research Projects Agency (DARPA) has embarked on a $50-million initiative known as "Human ID at a Distance," a major component of which is facial recognition.

Facial recognition can also have beneficial uses closer to home. For example, many parents would likely feel safer knowing their children's elementary school had a facial recognition system to ensure that convicted child molesters were not granted access to school grounds. Such a use, however, could point up one potential problem of facial recognition -- that people who have "paid their debt to society" may face heightened police scrutiny once they are identified in a public setting.

On the whole, however, biometric facial recognition systems offer advantages over other security measures. They are not invasive or even inconvenient. The system used at the Super Bowl was much less intrusive than a metal detector at a public building, or an inaugural parade checkpoint. In that sense, it helped to protect the privacy of individuals, who otherwise might have had to endure more individualized police attention.

The technological impartiality of facial recognition also offers significant benefits for society. While humans are adept at recognizing facial features, we are also susceptible to prejudices and preconceptions. The controversy surrounding racial profiling illustrates the problems that can result. Facial recognition systems, by contrast, do not focus on a person's skin color, hairstyle or manner of dress, and they do not recognize racial stereotypes. While there is a danger that the system may make an incorrect match, that danger is no more exaggerated than it is when traditional identification methods, such as comparing mug shots, are used.

While we must remain alert to potential abuses, we would be ill-advised to decry the technology's use under all circumstances. Instead, we should focus on monitoring what kind of information goes into watch list databases and what information is gathered, stored and disseminated. Options to consider include establishing legal measures to provide for responsible use; ensuring that citizens understand how the technology is used; monitoring government use through citizen oversight committees and review boards; and encouraging open, rather than surreptitious, use of the technology. The fear of potential but inchoate threats to privacy should not deter us from using facial recognition where it can produce positive benefits.

John Woodward is a senior policy analyst at RAND, where he works on biometric policy issues. The views expressed in this article are his own.

Biometrics: Will Digital Fingerprints, Iris Scans and Speaker Recognition Soon Replace Passwords and Personal Identification Numbers?

Biometrics is a high tech word for an old concept: how we go about recognizing one another. Biometric authentication uses automated methods based on physical characteristics or behavioral traits for human recognition. Examples of biometrics include iris and retina scanning, digitized fingerprints, hand geometry and speaker recognition.

Biometric authentication offers advantages over current security practices. Unlike keys and tokens, biometrics are never lost or stolen. Unlike passwords and PINs, biometrics cannot be forgotten. Because of their security, speed, efficiency and convenience, biometric authentication systems might soon become the standard for access control.


Biometrics Expert Delivers Lecture on Facial Recognition at RAND's Washington Office

Dr. James Wayman, a nationally recognized expert on biometric technologies, gave a lecture on biometric facial recognition at RAND's Arlington, Virginia office on March 15, 2001. Participants included RAND staff and representatives from the Department of Defense Biometrics Management Office and the United States Secret Service.


A PowerPoint slide presentation provides more information about the RAND team's research questions and findings.

High-Tech Human Identification Can Fight Terrorism

There is no foolproof solution. But biometrics could become a powerful way to maintain security

By John D. Woodward Jr.

This opinion article appeared in the Pittsburgh Post-Gazette on September 24, 2001.

As the nation begins to recover from this horrible tragedy, we need to dedicate our efforts to preventing any such terrorist acts in the future. While suicidal attacks can never be completely thwarted, we, as a nation, can take additional steps to counter them. We should explore many options.

Among them, we should examine the emerging technology of biometrics. While there is no easy foolproof technical fix to counter terrorism, biometrics might help make America a safer place.

Biometrics uses a person's physical characteristics or personal traits for automatic, nearly instantaneous human recognition. Digitized fingerprints, voiceprints, iris and retinal images, hand geometry and handwritten signature are all examples of characteristics that can identify us. While biometric technologies may seem exotic, their use is becoming increasingly common. Earlier this year, MIT Technology Review named biometrics as one of the "top 10 emerging technologies that will change the world."

There are several ways biometrics could be used to impede terrorism.

Currently many sensitive areas at airports are secured by badges and tokens. A person can swipe a pass and be given access to the runway, baggage loading areas and the airplanes themselves. Such measures are not wholly secure because badges and passes are easily forged, stolen and misplaced.

We can do better. For example, airline staff with a need to access sensitive areas of airports could be required to present a biometric like their iris to a sensor. From a foot away and in a matter of seconds, this device captures the person's iris image, converts it to a template,or computer readable format, and searches a database containing the templates of authorized personnel to attempt a match. A match confirms that the person is authorized to access a particular area.

This is not science fiction. Such a system is currently in place at Charlotte-Douglas International Airport.

The Immigration and Naturalization Service uses an Immigration and Naturalization Service Passenger Accelerated Service System where 65,000 enrolled, vetted international travelers voluntarily use hand geometry to verify their identity at ports of entry. This time saving use enables INS officers to spend more time on problem cases.


How Iris Recognition Works, by John Daugman

[JM Comments]
Description of algorithms used in comparing extracted IrisCodes. A match is typically accepted with a Hamming distance of 0.33 (estimated false match error rate of 1 in 3 million), but actual comparisons between typical normal-quality images exhibit Hamming distance of 0.24 (1 in 1015 false match error rate).

The fact that a Gabor wavelet model can be used for extracting pigmentation clusters is very significant. The need is for a robust metric for both textural variation and chromatic variation. A 2-D Radial Fourier transform will produce a texture metric, but the localization of color spots is a more complex task. Mapping iris image data to a wavelet model appears to provide the needed functionality.

The National Geographic Study The Afghan Girl

[JM Comments]
Details how iris identification was used to verify identity of woman after 17 year interval.

An interesting aspect is that the iris photos taken in 2002 (Figure 5) were vastly inferior in quality compared to the images from 1985 (Figure 2). They should have taken a Miles Research Digital Iris Camera with them for the iris photography. Possibly, the study protocol included intentionally degrading the recent images to demonstrate the robustness of the method. Alternatively, the images included in the document were not prepared for optimal appearance in digital format. In any case, the image conditioning was good at bringing out the details (Figure 7), and one can clearly see enlarged color spots in the right eye around 7:30, when compared to the 17 year-prior image (Figure 4).

Return to Links page