MANY people are unhappy about being filmed by CCTV in case the footage is used against them in some way. Now a surveillance technology company has come up with a method of scrambling the images of anyone in CCTV film who is not a suspect.
, based in San Francisco, says it should reassure members of the public who do not wish to be identifiable to police or lawyers, or even TV crime-stopper shows.
The technology uses 3VR鈥檚 recently patented face-recognition algorithms to home in on known faces in crowds. An image-scrambling algorithm then blurs the faces and bodies of those who are not of interest and encrypts the blur pattern so that no one but the operator of the technology can unscramble it (see picture).
Advertisement
鈥淎n image-scrambling algorithm blurs the faces of those in the footage who are not of interest鈥�
鈥淭his allows you to search for people on watch lists, for instance, but without capturing massive databases of innocent people,鈥� says Stephen Russell, 3VR鈥檚 chairman. The company aims to supply the equipment to banks and retail chains so they can analyse CCTV footage for known suspects who install card skimmers on ATMs, for example.
The idea is unlikely to satisfy all privacy advocates, since scrambled footage is still open to abuse. 鈥淎 safer approach is to record only when machine analysis detects something suspect- that way for 99.99 per cent of the time there is no recording of anything,鈥� says Mike Lynch of data-analysis company .
