Daniel Drake wrote:
> Hi Anthony,
>
> Please subscribe to the list so that we don't have to moderate your
> posts :)
Well... I did it but I'm not sure it worked. I haven't received any
confirmation or whatever. Should I retry?
> wxCover wrote:
>> A good news for those of you who have an aes1610 device.
>> I was working on a driver for it when Daniel contacted me a few days
>> ago.
>> I've adapted my code to fprint. I've attached a first version of this
>> driver based on the aes2501.
>> I've tested it and it seems to work quite well. You'll just have to
>> put the files in the driver dir and change the makefile etc. There's
>> also the authors file to update.
>
> Thanks a lot for working on this!
>
> I am very keen on including an aes1610 driver and doing a new release,
> a couple of trivial things I will change first:
>
> 1. ARRAY_SIZE no longer exists in the development tree, I switched to
> G_N_ELEMENTS from glib
Yes, sorry, my code is based on the 0.0.2 version. I should have worked
on the git version.
> 2. Tabs rather than spaces -- I know this is a holy war, but
> consistency is good and the rest of the code is tabs.
Ok, I'll do it now!
> 3. I'll assign a proper driver ID
>
> Before I do this, a couple of questions:
>
> Are the images always standardized correctly? By standardized, I mean
> the "finger_standardized.pgm" output file from the img_capture example
> should be dark finger ridges on a light background, and it should be
> natural upright orientation (no matter which way you scan your
> finger). Here is an example of a standardized image:
> http://www.reactivated.net/fprint/wiki/images/Fprint_demo_v0.2_verify.gif
This part works great: standardized output images are dark on light
background and the orientation is corrected. 2501 and 1610 gives the
same type of output images so I don't think there will be much problem
in this part of the code.
> How well is the image processing code working for this device? Is it
> reliably detecting when fingers do and don't match?
Well, I've spoke too soon in my last mail! I hadn't tried to enroll and
verify. The images I get looks nice BUT for the moment, I have problems
with enrollment and verification.
I think the main problem is that I'm not getting enough minutiae.
Usually, for a fingerprint that looks really nice, I get something like
10~15 minutiae. And if the image is not perfect (very often), the number
is less than 10.
Now, speaking about verification, I haven't manage to get it to match.
The scores are always low. I think this is due to the poor number of
minutiae.
I don't know how to improve the situation. Is there some tweaking we can
do with the algorithm which calculate minutiaes ?
I'll try to see if I can make it work better with my code.
> How many minutiae are being detected on a good scan?
> What kind of match scores are you getting for scans of the same
> finger, and scans of different fingers?
>
> To find the above info, configure libfprint with --enable-debug-log
> Then run the enroll example, and enroll a finger: it will tell you how
> many minutiae it found:
>
> fp:debug [fpi_img_detect_minutiae] minutiae scan completed in 0.233294
> secs
> fp:debug [fpi_img_detect_minutiae] detected 80 minutiae
> fp:debug [print_data_new] length=2404 driver=02 devtype=0000
> fp:debug [fp_enroll_finger_img] enroll complete
> fp:debug [fp_img_save_to_file] written to 'enrolled.pgm'
> Wrote scanned image to enrolled.pgm
> Enroll complete!
> Enrollment completed!
>
> 80 minutiae were detected.
>
> Then run verify and scan a finger, you'll see something like:
> fp:debug [fpi_img_detect_minutiae] minutiae scan completed in 0.233735
> secs
> fp:debug [fpi_img_detect_minutiae] detected 81 minutiae
> fp:debug [print_data_new] length=2404 driver=02 devtype=0000
> fp:debug [fpi_img_compare_print_data] bozorth processing took 0.021593
> seconds, score=57
> fp:debug [fp_verify_finger_img] result: match
> fp:debug [fp_img_save_to_file] written to 'verify.pgm'
> Wrote scanned image to verify.pgm
> MATCH!
>
> The match score was 57.
>
>> There's still some work to do on it. Especially gain settings.
>
> If it works for the device in your notebook, I'm happy to include this
> right away. Other drivers don't do gain calibration (yet). It's always
> good to get something out there if it is working to a degree.
>
> Thanks!!
> Daniel
_______________________________________________
fprint mailing list
[email protected]
http://lists.reactivated.net/mailman/listinfo/fprint
