2
COMMUNICATIONS OF THE ACM November 2002/Vol. 45, No. 11 23 T he Super Bowl is a sport- ing event deeply embedded in American culture. At the Super Bowl in January 2000, the police quietly began what may become a new tra- dition: the filming of specta- tors and the comparison of the taped images to a database of known criminals using advanced imaging technology. In such a way the police hope to be able to capture wanted crimi- nals who would otherwise escape detection. This is the latest event in a recent trend of filming public places in the interest of public safety. Since Sept. 11, heightened security awareness has led to the introduction of sweeping new measures of which increased video surveillance is only one compo- nent. Using advanced digital image technology, filmed images are scanned for the likeness of known suspects using a computer. And while law enforcement officials laud the benefits of such techniques in capturing criminals, civil libertarians claim the use of such cameras also presents a seri- ous intrusion on the right of the individual to pri- vacy and anonymity. To a certain extent, video surveillance has been used in public places such as subway stations and banks for decades. Until recently, such cameras would be monitored by a human being who would look for subversive or criminal activity. It was not possible to use a computer to draw conclusions regarding a recorded image, and as such this lim- ited the appeal to law enforcement officials of film- ing public places. The problems for a computer trying to identify a specific person from a photograph are many. Peo- ple change their manner of dress and their hair- styles from day to day. The photo the police have on record may have a criminal unshaven, with glasses, and not smiling, while the surveillance camera may film the same criminal neatly trimmed, with contact lenses, and laughing. To identify these two images as the same without the aid of human analysis poses a daunting technical problem that has only been recently addressed. There are several technological improvements that have made the Big Brother-type surveillance at the Super Bowl possible. For one, the quality of video images has improved vastly in recent years. As well, there have been theoretical advances in the development of complex mathematical algorithms that compare different images and look for matches. These algorithms have advanced to the point where they can still identify a person even if he or she has varied their appearance somewhat. And at the same time computers have become much more powerful and are more capable of per- forming the multiple complex computations these algorithms require. Together these developments have led to great advances in the field of so-called image processing used to identify a person in a photo. These new digital image processing techniques combined with increased computing power and image quality have recently eliminated much of the requirement for human involvement in identifying a face in a photo. Suddenly it is possible for a com- puter to scan an image of a crowd, and check to New Surveillance Techniques Raise Privacy Concerns LISA HANEY How much privacy are citizens willing to surrender in the interest of public safety? Viewpoint Boaz Gelbord and Gert Roelofsen

New surveillance techniques raise privacy concerns

  • Upload
    gert

  • View
    213

  • Download
    0

Embed Size (px)

Citation preview

COMMUNICATIONS OF THE ACM November 2002/Vol. 45, No. 11 23

The Super Bowl is a sport-ing event deeply embeddedin American culture. At

the Super Bowl in January2000, the police quietly began

what may become a new tra-dition: the filming of specta-

tors and the comparison of thetaped images to a database of known criminalsusing advanced imaging technology. In such a waythe police hope to be able to capture wanted crimi-nals who would otherwise escape detection. This isthe latest event in a recent trend of filming publicplaces in the interest of public safety. Since Sept.11, heightened security awareness has led to theintroduction of sweeping new measures of whichincreased video surveillance is only one compo-nent. Using advanced digital image technology,filmed images are scanned for the likeness ofknown suspects using a computer. And while lawenforcement officials laud the benefits of suchtechniques in capturing criminals, civil libertariansclaim the use of such cameras also presents a seri-ous intrusion on the right of the individual to pri-vacy and anonymity.

To a certain extent, video surveillance has beenused in public places such as subway stations andbanks for decades. Until recently, such cameraswould be monitored by a human being who wouldlook for subversive or criminal activity. It was notpossible to use a computer to draw conclusionsregarding a recorded image, and as such this lim-ited the appeal to law enforcement officials of film-ing public places.

The problems for a computer trying to identifya specific person from a photograph are many. Peo-ple change their manner of dress and their hair-styles from day to day. The photo the police haveon record may have a criminal unshaven, withglasses, and not smiling, while the surveillancecamera may film the same criminal neatlytrimmed, with contact lenses, and laughing. Toidentify these two images as the same without theaid of human analysis poses a daunting technicalproblem that has only been recently addressed.

There are several technological improvementsthat have made the Big Brother-type surveillance atthe Super Bowl possible. For one, the quality ofvideo images has improved vastly in recent years.As well, there have been theoretical advances in thedevelopment of complex mathematical algorithmsthat compare different images and look formatches. These algorithms have advanced to thepoint where they can still identify a person even ifhe or she has varied their appearance somewhat.And at the same time computers have becomemuch more powerful and are more capable of per-forming the multiple complex computations thesealgorithms require. Together these developmentshave led to great advances in the field of so-calledimage processing used to identify a person in aphoto.

These new digital image processing techniquescombined with increased computing power andimage quality have recently eliminated much of therequirement for human involvement in identifyinga face in a photo. Suddenly it is possible for a com-puter to scan an image of a crowd, and check to

New Surveillance TechniquesRaise Privacy Concerns

LISA

HA

NEY

How much privacy are citizens willing to surrender in the interest of public safety?

Viewpoint Boaz Gelbord and Gert Roelofsen

24 November 2002/Vol. 45, No. 11 COMMUNICATIONS OF THE ACM

see if a particular person is in that crowd. Thepotential of this technology in assisting the policein catching criminals is attracting the attention oflaw enforcement officials. This in turn raises thepotential that a vast video surveillance networkcould monitor a wide range of our public lives.

In such a scenario, a centralized police computercould serve as a collection point for the videoimages. By continuously checking the recordedimages against a database of known criminals, the

computer could alert police when it detected amatch. In the future, such a computer could alsoanswer more complex questions of the sort: WasMr. John Doe in the airport in the month of April?By comparing a photo of Mr. Doe with the digi-tized video recordings taken at the airport, thecomputer could then respond to this question.

Though this technology has yet to fully mature,the ability of computers to answer such questions isleading police and other organizations to installdigital video cameras in a number of public places.This raises the complex issue of how much privacycitizens are willing to surrender in the interest ofpublic safety.

There is little doubt that new video technologiescan provide a powerful new tool in capturing dan-gerous criminals. For this reason, society hasalready become largely accepting of video surveil-lance at ATMs, government buildings, and otherareas where security is a prime concern. To many,however, the recent deployment of video surveil-lance cameras in random public places is discon-certing. And as the technology of digital imagingmatures further, the deployment of such cameraswill only become more widespread.

There are other troubling issues related to videosurveillance that privacy advocates would like to seeaddressed. Where are the filmed images stored, and

who has access to the stored images? Video surveil-lance over the Internet is already a reality, so pre-sumably an enterprising criminal could gain accessto the recorded images and then use them forextortion. Also, the concept behind the video sur-veillance at the Super Bowl relied on comparing animage of the crowd with a database of photos ofknown criminals. Could this lead to the policemaintaining a database with the photos of all citi-zens, criminals and innocents alike? And what

would happen if the informationin such a database fell into thewrong hands?

At this point it seems thattechnology is moving faster thanpublic awareness and is outpacingthe public debate. As well, therush to introduce new security

measures in the wake of the events of Sept. 11 hasovershadowed privacy concerns. It may indeed beinevitable that the public will accept that it may befilmed at any point at any place, and that main-taining privacy will involve either a walk in thewoods or wearing a mask. Civil libertarians, how-ever, insist that any such acceptance be preceded bya vigorous public debate on the subject, in whichsociety considers the delicate balance betweenmaintaining law and order, and preserving the indi-vidual’s right to privacy.

In the myriad of technological changes currentlyfacing society and the host of accompanying ethicalquestions, such a debate has yet to seriously takeplace. Furthermore, the public’s current willingnessto sacrifice many aspects of privacy in order tocombat terrorism may mean the debate on videosurveillance will be put on the back burner. Thedanger is that by waiting too long there may be ade facto acceptance of the loss of privacy that willlater be difficult to change.

Boaz Gelbord ([email protected]) is an information securityconsultant at Royal Dutch Telecom (KPN) in the Netherlands.Gert Roelofsen ([email protected]) is employed by theDutch Ministry of Internal Affairs.

© 2002 ACM 0002-0782/02/1100 $5.00

c

Viewpoint

As the technology of digital imaging matures further, the deployment of such cameras will onlybecome more widespread.