Re: idea: brinworld meets the credit card

2003-07-11 Thread Adam Lydick
You might find facecerts interesting.

http://www.computer.org/proceedings/dcc/1896/18960435.pdf

This is more for face-to-face checking, however.

For your remote scenario some sort of one-way hash to verify the image
might be intersting. It would have to allow for fuzzy matching after
hashing (for obvious reasons). I think this just raises the bar a tiny
bit though, as an attacker could stalk their victim before stealing
their card to get an idea about what appearance to forge. (or capture
webcam traffic before lifting the card / identity info)

Cheers,

Adam Lydick

On Tue, 2003-07-08 at 12:16, Major Variola (ret) wrote:
 Authentication is Something you have / know / are.
 
 A simple plastic credit card + PIN provides the first
  two,
 including a photo provides the third something you are.
 A face is more often checked than the readily forgable
 signature, in live authentication.
 
 But as cameras become ubiquitous
 (e.g., in cell phones) some extra security could be obtained
 for *remote* authentication by sending a trusted photo of the
 account holder plus a live picture of the card user.
 
 A picture glued into the card could be forged, but a
 smartcard (with more data area than a magstripe)
 could include a picture of the account holder,
 so a thief has no idea what to look like.  But the vendor can
 check the encrypted smartcard face to the face on the phone
 or webcam.  For high-value remote transactions, where you
 pay someone to check faces, this might be viable in a few years.
 In a few years after that, machines might be able to check faces
 more cheaply, as reliably.
 
 The live face-check with embedded digital photos is already standard
 practice
 on high-security building-entry cards (and passports?),
 with the guard comparing the card-embedded face to the one before him.
 Ubiquitous cameras will bring that face-check to remote transactions,
 reducing cost due to lower fraud.
 
 Thoughts?



Re: idea: brinworld meets the credit card

2003-07-08 Thread Eric Murray
On Tue, Jul 08, 2003 at 12:16:36PM -0700, Major Variola (ret) wrote:
 Authentication is Something you have / know / are.

[..]

 A picture glued into the card could be forged, but a
 smartcard (with more data area than a magstripe)
 could include a picture of the account holder,
 so a thief has no idea what to look like.  But the vendor can
 check the encrypted smartcard face to the face on the phone
 or webcam.  For high-value remote transactions, where you
 pay someone to check faces, this might be viable in a few years.
 In a few years after that, machines might be able to check faces
 more cheaply, as reliably.
 
 The live face-check with embedded digital photos is already standard
 practice
 on high-security building-entry cards (and passports?),
 with the guard comparing the card-embedded face to the one before him.
 Ubiquitous cameras will bring that face-check to remote transactions,
 reducing cost due to lower fraud.
 
 Thoughts?

How does it allow the merchant to view the picture
while preventing the thief from doing so?

Saying it's encrypted is, at best, sweeping a very large
problem under a small rug.  Who holds the key?  How
does the card or the user authenticate a real merchant vs.
a thief posing as a merchant?

Those are the hard problems.  No one in biometrics
has yet been able to solve them in a general way.

Eric



Re: idea: brinworld meets the credit card

2003-07-08 Thread Morlock Elloi
 Those are the hard problems.  No one in biometrics
 has yet been able to solve them in a general way.

And the merchant example is the wrong application.

The merchant doesn't care WHO you are - that's a false premise.

Merchant cares if you can pay. Now, that's a completely solvable issue.

Of course, we know who and why is trying to misrepresent this.

All other applications of biometrics boil down to threatening with punishment
(we know who you are, behave or else ...) - and then the biometrics ceases to
be in the interest of the eyeball holder. Even granting door access to
employees fits this category. You don't let any qualified mathematician
willing to work to enter the lab - you let in only those that you know where
they live, have signed contracts with them, etc.



=
end
(of original message)

Y-a*h*o-o (yes, they scan for this) spam follows:

__
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com