|
A meeting with the professor September 2007
Professor John Daugman is the father of biometrics based on irisprints. He gave a lecture in Cambridge, England, on Saturday 22 September 2007, Identity in the Eye. Much of the lecture followed his 2001 paper, The importance of being random: statistical principles of iris recognition, which makes the case for irisprints. He now has a database of 200 billion matches, up from 9 million in 2001. He has more examples of where irisprints are being deployed -- eBorders/IRIS, the UAE, Iraq, Afghanistan and the US prison service. And he has some impressive (ROC) data showing that whereas other biometrics have a curve -- if false matches go down, false non-matches go up, and vice versa -- irisprints just have a point, i.e. irisprints are the best biometric.
One person asked about post mortem identification. Professor Daugman believes at the moment that the iris could be used for identification purposes for up to three days after death. Spoofing is a danger which worries him about the use of irisprints -- the technology largely relies on co-operative subjects. The question arose at what distance irisprints can be reliably taken. Currently up to about 1m, according to Professor Daugman, there may be a natural limit of about 10m and his recent suggestion at a lecture in San Francisco that when the Hubble Telescope is de-commissioned it should be pointed at the earth to watch everyone was a joke. There aren't many jokes in this field and he managed to use the other one* when we were talking outside after the lecture. My daughter joined us. "Lovely," he said, "she has your eyes". On that point, one of the strengths claimed for irisprints, like fingerprints, is that they are not genetically determined: identical twins have different irisprints; and your two irisprints are different. Another strength claimed is that they seem to remain recognisably constant throughout post-natal life. (The colour of the iris is genetically determined but nothing else, none of the "arching ligaments, furrows, ridges, crypts, rings, corona, freckles, and ... zigzag collarette" that are discerned in the iris. Irisprints are monochromatic, taken in the infra red band, outside the visible spectrum, just to the left past red, so colour doesn't come into it.) One warning from the Professor -- do not rub peanut butter into your fingers. Like hand cream, this can adversely affect the ability of scanners to read your fingerprints. Was that 10%/39% failure to acquire rate a serious problem for biometrics based on irisprints? Not to mention the 4%/9% false non-match rate. I will continue to think so unless and until there is good evidence to the contrary. ---------- David Moss has spent five years campaigning against the Home Office's ID card scheme.
|