[fprint] Problems with pam_fprint

[Daniel Drake]

João Miguel Roque wrote:
> This solves the problem for regular users to use the lib, but when
> running pam_fprint_enroll, it scans the finger, but then it can't save
> the file and gives the following error:
> 
> 	Scan your finger now.
> 	Enroll complete!
> 	Enrollment completed!
> 
> 	fp:error [fp_print_data_save] couldn't create storage directory
> 	Data save failed, code -1
> 
> Where is the lib saving the files so that we can change permissions?

~/.fprint

> Other thing is that the scan at least with the AES1610 doesn't have a
> good quality and maybe this is not only with the quality of the sensor,
> but also with the speed one must use when scaning the finger. If you
> scan too slow, it gives a segmentation fault. Don't know the internals
> of the program, but if possible, try to slow a bit the scan so that we
> can avoid this segmentation fault. Other thing could be to add in the
> demo a param that will do that, we the users will check the best speed
> for our sensors and then you could implement that speed in
> pam_fprint_enroll.

No matter how slow the user scans their finger there should never be a 
segfault. Can you run it through gdb and get a backtrace?

Anthony, can you reproduce this?


As for the quality problems, can you elaborate? Do you mean that the 
print comparison simply isn't working brilliantly (that's a known issue 
for this particular device), or do you mean that the images that come 
back actually look bad in the first place? If so, in what way do they 
look bad?

The example images that Anthony posted looked good quality and also 
enhanced well too, the issue seems to be that our image processing 
doesn't work brilliantly for such small scan images at the moment. 
(other sensors see a lot more of the finger)

I do not have this hardware myself so am not able to directly relate to 
any of this.

Daniel