Nick Wagers wrote:
> Print quality of enrollment prints: The NBIS library offers NFIQ, it
> scores the image quality. This could be used during enrollment to
> ensure that only the best fingerprints are obtained. It could also be
> used to present errors when fingerprints fail verification.

Already planned and Davidlohr Bueso has posted a patch which adds NFIQ. 
Once I have integrated this, the final step will be to add the 
functionality you describe.

> Device options: libfprint could set/get hardware options. One comes to
> mind, the U.are.U 4000/Microsoft Fingerprint reader has an encryption
> option, which allows the transfer between the device and the computer
> to be encrypted.

We do not know the encryption scheme so this would be useless - we 
cannot decode the encrypted images. Do you have details?

> PAM: I'm not sure how flexible PAM is but it would be nice to allow
> multiple tries during verification. I've had some fingers fail
> consistently (maybe I have ugly prints?).  Also, it may be useful to
> keep multiple good prints for PAM to check against and I see no reason
> why PAM authentication shouldn't allow for a match on any finger
> instead of a specific. Users can enroll only the fingers they wish to
> use.

There were reports that NBIS works better on some fingers than other 
(young vs old was discussed before).

pam_fprint was a quick proof of concept, I'd accept patches for a 
"retries" parameter.

As for comparing against multiple prints, it's not quite that simple 
because we cannot do that in the upekts driver (which supports a few 
very popular non-imaging devices). That said, Vasily K has announced 
that he is working on a more functional PAM module which may support 
identification for devices that can support it.

Thanks for the feedback.
Daniel

_______________________________________________
fprint mailing list
[email protected]
http://lists.reactivated.net/mailman/listinfo/fprint

Reply via email to