B
Boris Mohar
- Jan 1, 1970
- 0
Boris said:
Let them try that with a retinal scan lock....Boris Mohar said:
greysky said:Let them try that with a retinal scan lock....
Boris said:
Boris Mohar said:
Rene said:To many scripts on this page...
Why does anyone think a fingerprint system is safe ?
Because the FBI once claimed the (rolled!)
fingerprints are unique to 1 in 1E9 ? This claim has
never been proven, by the way. I once was involved :
http://www.ibrtses.com/projects/fingerprint.html
The results weren't that good, and solutions tricky.
Frithiof Andreas Jensen said:Here is a HowTo:
http://www.diva-portal.org/liu/abstract.xsql?dbid=2397
Whats really neat is that Marie Sandström used a PCB to provide the
mould for the fake fingerprints (and of course that we need wimmen in
enginerring)
ian field said:Talking of PCBs, and its likely less of a problem with the relentless
march of SMD, when I used to handle a lot of PCBs my fingerprints were
unreadable! - I'd have stood less chance of gaining authorised access than
the most amateur hacker!!!
Rene Tschaggelar said:To many scripts on this page...
Why does anyone think a fingerprint system is safe ?
Because the FBI once claimed the (rolled!)
fingerprints are unique to 1 in 1E9 ? This claim has
never been proven, by the way.
You're off by a few orders of magnitude, Rene. 88 orders, in fact. The FBI
claims 1 in 1E97.
Mike said:You're off by a few orders of magnitude, Rene. 88 orders, in fact. The FBI
claims 1 in 1E97.
Isn't that amazing? What's even more amazing, though, it that they think
they _have_ proven it.
http://www.newscientist.com/article.ns?id=dn4611
Epstein, Robert, "Fingerprints Meet Daubert: The Myth of Fingerprint Science
is Revealed," So. Cal. Law Review, vol 75:605, 2002, p.630.
If your original number, 1 in 1E9, was correct, then (extending the birthday
paradox to 1E9 possible birthdays) in any city of 38,000 people, the odds
are greater than 50% that two people would have identical fingerprints. It
doesn't exactly inspire confidence in the system, does it?
If your original number, 1 in 1E9, was correct, then (extending the birthday
paradox to 1E9 possible birthdays) in any city of 38,000 people, the odds
are greater than 50% that two people would have identical fingerprints. It
doesn't exactly inspire confidence in the system, does it?
Rene Tschaggelar said:The last time I was involved, we would have been thrilled
to even be be able to verify 1e6 FAR @ 1e2 FRR or so.
It looks far simpler than it really is hands on.
Mike,Mike said:I agree, Rene. I'm not a fingerprint expert, but a brief review of the
fingerprint literature is devoid of the keywords I thought I would find.
There's no mention of noise, distance, noise enhancement, error rate, or
anything else that a communication engineer would expect to see. I was truly
amazed to see that the fingerprint community believes that the error rate of
the fingerprint identification process is zero.
Latent prints are often highly filtered to "enhance" the high frequency
components so an identification can be made. Even though that should lead to
errors, the fingerprint community seems to be completely ignorant of the
effects of high pass filtering on noise.
-- Mike --
there are basically two communities. One is the law
& law enforcement community and to them a fingerprint is
error free, and their view is little to not opposed.
And then there is there is the community of automated
fingerprint authentication devices. They have numbers
such as false acceptance ratio, false rejection ratio
traded against each other.