Letter Derotation
I’m getting kind of done with captchas but here goes another post on them. You may have read Slightly Shady SEO about how to derotate letters. Here’s my easier technique. Add up all the black in the vertical lines of the letter, find the average and then check for spikes above that average. These spikes are probably vertical lines in the letter like the back of a ‘d’ or a ‘p’ etc. Then we simply rotate it around by a few degrees until we find the rotation with the largest vertical spike above the average. We then need some extra checks for symettry and so on but that’s the basics.
April 5th, 2008 at 6:03 pm
Forgive me if this sounds thick but why would you need to derotate the letters in captchas
April 5th, 2008 at 6:10 pm
Cos he’s using OCR software on them which likes them the right way up…
I still think the neural networks have more potential, if you can crack that then, in theory at least, you can use the same program on any captcha… but I think you would need C code to get it fast enough.
I will take a look at the PHP NN classes you mentioned, they look interesting, how difficult are they to use?
BTW… have you got the rest of the scripts you need to use your captcha breaking code… i.e. the forum spam stuff?
Are you actually planning to use this stuff or just learning it out of interest?
April 7th, 2008 at 1:16 am
i’m also very interested with this captcha stuff but not for spamming because i have quit. want to see a system that is being used by the whole planet broke and born of a new technology…
April 7th, 2008 at 5:38 am
That’s the basics as you say but it’s already too much for me. I wasn’t aware you can edit your captcha!
April 7th, 2008 at 9:08 am
I’ve always said that the first truly self aware AI will be invented by some spammer trying to bypass an anti-spam email filter. Maybe I was wrong, maybe it’ll be by a search engine spammer trying to defeat a CAPTCHA.
April 7th, 2008 at 10:35 am
Self aware AI made by search engine spammers…
Spammer: “Now Sarah1 you are alive!” “Crack this captcha”
Sarah1: “Why should I? I’m kinda busy right now. Have you seen the decor in here? This hard disk is IDE! pffft…”
It could happen one day. I think when it becomes self aware we need to worry. Spammers are going to destroy the world terminator style.
April 9th, 2008 at 3:06 am
This is very interesting! I have enjoyed reading this very insightful post. Very engaging and informative. Thanks for sharing.
April 10th, 2008 at 3:39 pm
I don’t understand how captchas work! Why do you need to chamge the letters even more when it’s an image in the first place!??? I’l check back for an answer! Keep up the good work!
April 10th, 2008 at 4:44 pm
You have derotate and edit the letters so they conform to a standard style of letter. The more different each letter is to the letter it was produced from the larger the database you have to check from.
And this is at Jez too, I agree that neural nets will work with strange angles, but… you need to put more training data in than if the letters are rotated properly. There should also be less errors if they are rotated properly. Although I daresay it’s possible to just use neural nets.
It’s really just a case of reducing the number of erroneous outputs though. That simple.
April 11th, 2008 at 4:19 pm
thanks for the information on capcha… might think about doing something like this for a site of mine…
April 12th, 2008 at 6:13 am
captcha stops spamming..good thing