Hi Daniel,

Daniel Drake schrieb:
> Roman Schließmeyer wrote:
> I don't really understand your mail - first you say that 
> verification/pam_fprint works fine, then you say you can't get it to 
> accept a fingerprint?

Sorry for confusion. What I meant was, that libfprint and pam_fprint 
compile and start clean. Also the device gets properly detected and 
scans the fingers e.g. in fprint_demo.

My problem was I could not verify a fingerprint against an enrolled one, 
but I was just testing around again and experienced that the matches are 
pretty good when using exactly the middle of the fingertip (where the 
most "curls" are).

> Anyway, if you're having false rejection issues, then you should look at 
> the images and examine their appearance. Look for how smooth they are, 
> how similar they appear, how much of the finger they cover. Also if you 
> use fprint_demo you can ask it to plot the minutiae - see how good a job 
> it does at that. Finally you can enable libfprint debugging and see the 
> bozorth3 match scores for when it compares to the default threshold of 
> 40 (40+ = match).

Thanks for the hints - I get definetly more minutiae counts when using 
the fingertip more centered on the reader - I think therefore the 
matches are also getting better.

Greetings,

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

Reply via email to