[fprint] EgisTec ES603 Error -22

[Patrick Marlier]

Hi Martin,

On Thu, Mar 20, 2014 at 9:52 PM, Martin Hejnfelt <martin at duffman.dk> wrote:
> Hi,
>
> I'm working with a EgisTec ES603 (1c7a:0603). I am using the current
> HEAD from anarsoul git repository.
>
> When I try to enroll or identify (with async methods), sometimes it
> works, but most of the time I get a result of -22 (EINVAL) in the
> callback. I tried debugging it, and it seems the error comes in the
> sanitize_img function (imgdev.c line 93) from a check of the image
> height (line 106-111, error comes in 110). By enabling logging, I can
> see that during an enroll sequence, my callback function is called twice
> from one call. The first call is good, where I get an FP_ENROLL_PASS
> swipe and the function exits as it should. It then immediately gets
> called again, however this is the one that fails with error -22.
>
> I then enabled logging in the library, and it seems that it detects the
> finger twice or something, and then removes some empty lines that end up
> with a blank image. Here's some debug output to gaze upon:


Thanks for the debug output ! :)


> fp:debug [fpi_imgdev_report_finger_status] finger on sensor
...
> etes603:debug [process_remove_fp_end] Removing 354 empty lines from
> image
> etes603:debug [process_remove_fp_end] Removing 0 empty lines from image
> fp:debug [fpi_img_new] length=163840
> etes603:debug [m_capture_state] Sending the raw fingerprint image
> (0x640)
> fp:debug [fpi_imgdev_image_captured]
> fp:debug [fpi_img_detect_minutiae] minutiae scan completed in 0.057886
> secs
> fp:debug [fpi_img_detect_minutiae] detected 97 minutiae


Indeed first capture looks fine even though 354 lines are removed from
the capture. How fast did you swipe? Did you try fprint_demo to check
the quality of fingerprint?


> etes603:debug [process_remove_fp_end] Removing 498 empty lines from
> image
> etes603:debug [process_remove_fp_end] Removing 496 empty lines from
> image


Here the second fingerprint seems empty... Not sure what happened
here. Is it easy to reproduce?

I think the best way to find the problem is that you try fprint_demo
and you send me debug output and resulted fingerprint images.

Thanks!
--
Patrick