Robert Cummings wrote: > On Thu, 2006-05-11 at 13:48, tedd wrote: > >>At 12:11 PM -0400 5/11/06, Robert Cummings wrote: >> >>>On Thu, 2006-05-11 at 11:47, tedd wrote: >>> >>>> At 9:28 AM +0300 5/11/06, Dotan Cohen wrote: >>>> >Hey all, it is possible to parse capcha's in php? I'm not asking how >>>> >to do it, nor have I any need, it's just something that I was >>>> >discussing with a friend. My stand was that ImageMagik could crack >>>> >them. She says no way. What are your opinions? >>>> > >>>> >Thanks. >>>> > >>>> >Dotan Cohen >>>> >http://what-is-what.com >>>> >>> >>> > Of course -- it's trivial. >>> >>>> All images can be broken down into signals and analyzed as such. If >>>> you have any coherent data, it will show up. If it has to conform to >>>> glyphs, it most certainly can be identified. >>>> >>>> You want something that's not trivial, take a look at medical imaging >>>> and analysis thereof. >>> >>>Extracting passcodes from captcha text is not what I'd call trivial. >>>It's one thing to pull trends out of an image, it's quite another to >>>know that a curvy line is the morphed vertical base of the capital >>>letter T. Similarly knowing that the intensity of red in an area is >>>related to the existence of some radioacive tracer agent, isn't quite >>>the same as knowing that the curvy letter T might be red, yellow, green, >>>yellow blended to green,. etc etc. The human eye and brain are amazing >>>accomplishments, and while someday we may match their ability in code, I >>>don't think it's this year. >> >>We've been doing edge detection, noise suppression, data analysis, >>and OCR for over 30 years. While it may not be obvious, it's still >>trivial in the overall scheme of things. The bleeding edge is far >>beyond this technology. > > > Edge detection, noise suppression, and data analysis don't quite equate > to recognition. Also 30 years of OCR still requires that the sample be > good quality and conform to fairly detectable patterns. If this is so > trivial, I await the release of your captcha parser. The spammers would > probably pay you millions for it. Where exactly is this bleeding edge, > and where can I read more about it? I think you're quite wholeheartedly > being naive about the complexity of visual recognition. Prove me wrong. I also agree most are breakable. I've done a very small amount of character recognition processing and most of the captcha's I've seen would be breakable. The ones that look hard such as the bugs.php.net captcha, I end up getting wrong about 1/3rd of the time. There is a substantial difference between standard OCR and captcha breaking. With OCR you need to get it right 99% of the time, with a captcha if you can get it one time in 1000 you can still get into a website several times a second. Really though, there are easier ways to do it. My favorite story was a small free porn site that required you to enter a captcha to get in. They were taking the captcha's they needed to break and getting horny teenagers to do the recognition phase for them. -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
