Pavel - thanks for the reply


upekts does *not* support identification, due to a difference in its inner

workings (it uses the hardware itself to do the matching, and the part of the

protocol used for identification or image functions is not known, although the

device itself is capable of such functions with windows)



MG: This is what I feared :(



A follow up question:  upekts returns a 200 byte identification string.  Has

anyone tried using these for comparison purposes?



I notice from Daniel's driver page that he has proved that the 200 byte

string is a sufficient unique ID, because he uploaded to a second device

and verify worked.   I am not quite sure if this implies that we can

use these for internal comparison though?



any other image-based sensor (i.e. sensor with driver that actually uses

libfprint for matching) will do identification, though



don't quite understand what you mean by "image" sensor, as swipe sensors also

provide a result as an image, or rather a series of image that is used to

reconstruct the fingerprint.

the difference between touch and swipe sensors is that swipe sensors are

smaller, capable of providing longer (and variable length) images, somewhat

more difficult to use, and the driver is a little bit more complicated (due to

the image reconstruction part).

one more quirk of touch based sensors is you leaving the actual fingerprint

right on the device, but that is probably irrelevant



OK, I was a bit confused in my terminology.



I've been looking at devices on Upek's website.  They have both the touch

and swipe sensors on their site, but nothing about how you would decide

which to use.



My interest is in questions of accuracy and user ease-of-use.  It seems

above that you're suggesting that the non-swipe devices are better in

this regard?



On the other hand, I've read that the upek sensors with hardware ID chip

are a lot faster in use.



So the jury is out I guess!



Martin Green
_______________________________________________
fprint mailing list
[email protected]
http://lists.reactivated.net/mailman/listinfo/fprint

Reply via email to