Originally Posted by Lucy
I assume the data requirements were onerous for those things back in the 80's.
It wasn't the image that was being tested.
The signature got characterized with a few vectors, and they weren't terribly huge iirc. The R&D guys didn't give my dad the total story, but signature recognition was done via pressure gradients. It has nothing to do with what it looks like .... that's why none of the forgers were successful.
I do bet they used Hidden Markov Models... great stuff for that sort of thing.
Now offering online seminars, live seminars, and everything else under the sun and over the moon for actuarial exams.