Unisys eyes moving targets to identify
- By Trudy Walsh
- Nov 19, 2008
The eyes are the windows to the soul, the saying goes, but they also might be the window to identifying large numbers of people moving through airports or other hubs. Unisys is working with governments and other organizations to develop a biometric technology named 'iris on the move.'
Iris recognition technology works with monochrome images of the eye, said Mark Cohn, vice president of integrated security programs for Unisys. The company is taking an approach that 'lets you take iris images at greater distances,' Cohn said. 'It can find the face of a person as they move at their normal walking pace.'
Unisys is working with Global Rainmaker Inc. to deploy the HBox, a device that can be mounted over a doorway to perform iris recognition at a distance. HBox can acquire an image of a person's iris as they move toward the door, and the image can then be used for biometric identification and authentication, Cohn said.
With HBox, about 20 to 30 people per minute can walk through the HBox doorway and have their iris image captured from a one- to two-meter distance, Cohn said.
Iris recognition has striking accuracy advantages over other biometric technologies, Cohn said. It delivers a low false match rate when used to verify that an enrolled participant is who he or she claims to be.
In contrast, fingerprints work for populations up to tens of thousands, Cohn said, but beyond that number, the technology runs into matches that may not be unique. About one in 10,000 fingerprints is a false match, he said. With fingerprints, security operations have to add a second biometric. With iris recognition, organizations can use the technology by itself, Cohn said. Iris technology might find a false match in tens of millions of images, he said.
And iris patterns don't change from childhood to old age, so enrollment may be once for a lifetime, Cohn said.
Iris recognition technology does have a few drawbacks, Cohn said. Compared with fingerprint or facial recognition, iris capture devices are much more expensive, the systems are harder to integrate, subjects have to be more cooperative and user training is more difficult.
To enroll, subjects must stand still near a camera for two to three seconds, sometimes holding their eyes wide open. Even if subjects cooperate, acquiring an acceptable image isn't guaranteed, Cohn said.
Other disadvantages of iris recognition Unisys cited include vendor algorithm/template interoperability; lack of usefulness in forensic analysis because iris images aren't left at crime scenes; few implementations with large databases; and public perceptions that confuse iris recognition with more intrusive retina scanning technology.
Trudy Walsh is a senior writer for GCN.