Hi Dan, Thanks for confirmation, I thought it was just AES3500 performs badly.
Is it the matching approach (i.e. minuates detection and matching in NBIS) that poses a limitation on verification of these kind devices? Is the Windows driver using a completely different approach on verification? Sorry, my experiences are very limited, would very much like to get your advice! Thanks! best, -- jv ♫ m: (+86) 156 5238 4808 On Wed, May 8, 2013 at 10:03 PM, Daniel Drake <[email protected]> wrote: > El 08/05/13 02:54, Juvenn Woo escribió: > > Dear all, >> >> I'm submitting the AES3500 driver for peers to review. >> >> The AES3500 is a pretty old fingerprint device produced by AuthenTech. It >> possesses a *press-typed* sensor of dimension in 128x128. Which is >> similiar to AES4000, except the later is in 96x96. >> >> The driver is a deriative work of Daniel Drake's AES4000 driver, with a >> few parameters tuned for AES3500. While there seems no problem scanning >> fingerprint images with it, the verification rate is pretty low at the >> moment. >> > > The AES4000 driver performs badly as well. As you've commented, the likely > problem is that only a tiny area of the finger is sampled. > > I imagine the windows driver (where the sensor seems to perform OK) has a > fingerprint processing engine heavily tuned to small amounts of data. It's > probably quite insecure as a result. > > Thanks for the contribution! > > Daniel > > > >
_______________________________________________ fprint mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/fprint
