[Fingerforce-devel] 0.4 release soon?

[Daniel Drake]

Hi Radek,

Radek Bartoň wrote:
> I did't know about fprint project before I started working on FVS. How old is 
> this project? I need to read deeply its wiki to get image about how are all 
> these projects (FingerForce, aes2510 kernel driver, aes2501 usbrunner and 
> FVS) related?

I've been working on it for almost 2 months but only released it 
recently. Apologies for the earlier silence here, I am doing it as an 
academic project too and had to implement the fundamentals myself before 
  opening it to the community. At this point I am running it like a 
standard open source project and will happily take reasonable contributions.

As for relationships with other projects:
  - fprint replaces aes2510 kernel, aes2501 userspace, dpfp, thinkfinger,
    and whatever other fingerprint projects there are. It includes it's
    own userspace drivers internally and does not depend on any of those
    other projects.
  - FingerForce is (I think) Debian's effort to integrate fingerprint
    scanning into their distribution. fprint and it's subprojects will
    hopefully reach the stage where it can fulfill those requirements
    on it's own, and I hope they'll be considered as the Debian solution
    here. I will also be working with other distros and upstreams to get
    fprint integrated elsewhere.
  - fprint includes NBIS which can be seen as a replacement for FVS.
    I've experimented with both (but not your latest FVS work), and NBIS
    is much much faster and more accurate. It is used in the USA in the
    FBI and the DHS so it is quite proven too :)

That said, NBIS is not working out too well for dealing with images from 
sensors which see smaller areas of the fingerprints. In other words, the 
matching results are good-but-not-great for images with fewer minutiae 
(whereas results are *excellent* for images with a decent amount of 
minutiae). I'm confident that it will be possible to make improvements 
to NBIS to increase accuracy for smaller images, but I'm not sure how 
large this task would be. I don't really have much familiarity with the 
NBIS internals yet.

Competition is always good in these areas and I would definitely 
consider alternatives if something comparable to NBIS is developed, so 
if you are interested I would encourage you to continue working on FVS. 
Or, if NBIS sounds interesting to you, help working on the NBIS 
internals is also much appreciated.

Daniel