Re: [xcbf][Fwd:I-DACTION:draft-nagpal-biometric-digital-signature-00.txt]

[Alessandro Triglia <sandro@oss.com>]

At 2002/05/16 18:46 +0100, John Larmouth wrote:
>Alessandro Triglia wrote:
> >
> > At 2002/05/16 16:01 +0100, John Larmouth wrote:
> > >This is quite interesting!  He admits it is not possible yet!
> >
> > This proposal is very far from reality.  The probability of getting two
> > identical biometric samples from the same person are very low, for any
> > biometric technique,
>
>No, I don't think so.  For hand geometry, I think we could be close to
>doing what he wants, and possibly for fingerprints, where they count
>cusps and whorls etc.


John,

We are talking about taking measures of real-world objects.  No two images 
of a fingerprint will be identical, especially when taken at different 
times and on different systems.

The representation of a fingerprint usually includes (see ANSI/NIST-ITL 
1-2000) the representation of a vector (cartesian coordinates of the 
origin, phase angle and length) for each of the identified minutiae ("ridge 
ending", "bifurcation", etc.).  These measures can vary from sample to 
sample.  It is even possible that a "ridge ending", for example, is 
identified as such in a sample but not recognized in another sample, so the 
very list of minutiae (let alone their coordinates) may vary from sample to 
sample.

Alessandro Triglia
OSS Nokalva



>But for retina and iris stuff, I agree with you -
>we are nowhere near being able to do it.
>
> > OSS Nokalva's biometrics toolkit will support all the techniques of
> > encryption, digital enveloping, digital signing, and MAC-based
> > authentication that are defined in X9.84.
>
>My biggest problem with both X9.84 and the BIR is the failure to
>separate the payload from the biometric data.  But that is a different
>discussion.
>
>But in summary, I said in my original:
>
> > >We should not waste time on this.
>
>And I think that still stands.
>
>John L
>
>----------------------------------------------------------------
>To subscribe or unsubscribe from this elist use the subscription
>manager: <http://lists.oasis-open.org/ob/adm.pl>