Re: Biometrics

[Eduardo Kienetz <eduardok () gmail com>]

On 7/13/05, Ansgar -59cobalt- Wiechers <bugtraq () planetcobalt net> wrote:
> On 2005-07-12 Eduardo Kienetz wrote:
> > On 7/12/05, Ansgar -59cobalt- Wiechers <bugtraq () planetcobalt net> wrote:
> > > 1. With biometrics you always have to find a balance between false
> > >    accepts (wrong person get's access) and false rejects (valid user
> > >    doesn't get access).
> > > 2. Fingerprints can be easily forged [1], and people leave their marks
> > >    around everywhere they go.
> > > 3. How will you handle a biometric token (i.e. fingerprint), that gets
> > >    compromised? People usually have only ten fingers.
> >
> > Just a clarification here...
> > This is not a problem anymore... there are new fingerprint (even whole
> > hand) scanners that not only scan your finger/hand, but also measure
> > temperature/pulse (to make sure the hand is alive :).
>
> You haven't read the article I mentioned, have you?

The article only shows someone copying a fingerprint, but no tests are
made with a fingerprint recognition device. I wouldn't take
conclusions from that. Also, the fingerprint is very well 'printed' in
the bottle. Now, do you know of any studies with percentage of
fingerprints been acquired in such a good shape (or, from such
objects)? Note that I'm not opposing to your opinion, just adding more
details into it.
By the way, I've read many other articles besides that one.

> > Besides that if you use password-based auth, the "thief" would just
> > need to threat you that... for example he'll cut your finger if you
> > don't tell him the password... ;) etc.
>
> And you would consider this to be easier than getting someone's finger-
> print from e.g. a bottle or glass in a restaurant, because ... ?

I was comparing it to the fact of cutting someones finger, which was
pointed out by our colleague. Nothing else. I agree getting someone's
fingerprint from a bootle or glass is easier (although I maintain
citation above regarding the superficial article).

 
> > One could even combine the scanning of BOTH hands to authorize.
>
> That would not only fail to solve the inherent problem, but also reduce
> the pool of available tokens from 10 to 1.
True. Perhaps you got me wrong.

Best regards,

-- 
Eduardo  Bacchi Kienetz
LPI Certified - Level 1 & 2
http://www.noticiaslinux.com.br/eduardo/