OCR = Optical Character Regognition.
Although I dont know why you mentioned using OCR on those images since they are supposed to be designed specifically to fool OCR software so it cant correctly read it, and so only the human user can make it out.
True. The class I mentioned was written specifically for an online game that uses this type of "human-detection". What it does is filter out most of the quirkies put into the image to make them hard to decypher, do a remapping to a smaller b/w image and then perform a simple OCR. For the site it was built for, it's about 90% accurate. In the other 10%, the site won't accept the registration (in which case the script simply tries again).
I think using this technique (and doing it properly) will scare away most script kiddies though.
You don't have to stick with numbers and letters. A picture could also be send, while the server asks (...through jabber:x:data?):
"What do you see on the picture?" A) A child playing with a ball B) A horse C) A glass of milk
That to-fake-script will probably be a bit harder to write :p.
Mattias
_______________________________________________ jdev mailing list [EMAIL PROTECTED] http://mailman.jabber.org/listinfo/jdev
