Davidlohr Bueso wrote:
> OK,
> 
> Two new functions now, after separating the initial code:
> 
> float fp_img_get_quality_confidence(struct fp_img *img);
> return the img quality confidence score
> 
> int fp_img_get_quality_score(struct fp_img *img);
> returns the img quality score

That's more what I was thinking, thanks. I now see why you zipped it 
(too big for ML).

I hadn't twigged that your patch was to add NFIQ. Nice work! I was 
planning to do that at some point. This will form the basis of an 
improved enrollment mechanism in future (scan 2 or 3 times and pick the 
best image according to NFIQ's rating).

Are the NFIQ files straight copies or have you made changes to them? 
Which version of NBIS did these come from?

The patch includes some junk - modified Makefiles, .git stuff, etc 
(files not under revision control). Maybe time for you to read the git 
user manual? Hint: do your work, commit it, and use git-format-patch to 
get it into email form.

I assume calculating the quality is quite an intensive operation? In 
that case it should not be done twice if someone wants to get the 
quality and the confidence.

Your confidence documentation doesn't give any indication of how to 
interpret a non-error result.

Finally a question: when I looked at NFIQ before I gathered that it 
relied on some smallish text file being installed in a certain location. 
  How have you gotten around that? (or am I mistaken?)

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

Reply via email to