How does image verification work?

Let me answer this -

Let me answer this.

How would you differentiate a human & a computer on the internet? How would you differentiate a 'bot' (internet robots) from a real person? How would you prevent someone from running automated scripts which would create fake ids on email servers?

The answer comes from CAPTCHA - Completely Automated Public Turing test to tell Computers and Humans Apart. The technique implements a distorted image consisting of digits & letters which can be understood only by a human eye. The person filling out the registration form is asked to enter the alpha-numeric combination which appears in the image. Therefore, even if a script is run to fill the registration form; it will fail because the image verification will fail.

Of course, a new set of letters & numbers is selected every time. The reason is obvious, isn't it? We do have the same technique to prevent fake registrations.

I hope this answers your question. Got doubts? Let them in. 😀

-The Big K-

about internet robots

I'll try to explain the concept of internet robots. I am sure that most of us have heard about the search engine crawler. Googlebot or Yahoo Spider. These are typical examples of the internet robots which 'crawl' the webpage and update their indexes into a central database.

Internet robot is just a software code that does this magic. Millions of web pages are crawled by these bots everyday.

Another classic example of internet robots is 'Yahoo Helper' in Yahoo Messenger. Its is a software that reads incoming messages. Filters the keywords and then depending upon the keyword, they send back appropirate messages.

Jerry.

thanks

That was quite some information, big k & jerry ! More details on Turing Test please :smile: !

Crook

Thanx Biggie and Jerry!!!

I would like to add to that. The random images are auto-generated aswell in many cases, but not all. Surprising????

The images are a random combination of font, size, orientation, colour and location of each indvidual character. Once this combination is generated, the data pertaining to the image is saved from display memory i.e. what we see on screen and not from data memory, which is set of characters with attributes mentioned.

Why different fonts, colour.....? There are image recognition softwares which can identify text. We simply need to generate a very complex pattern which a human can understand but a software might not.

Does this mean that the techinique is fool proof??? Most of these images have different characters in different colours. A cue to isolate and identify each character. Then we can always know which character follows what.

To verify this, I used an image2text software and found that more than 70% of times the software actually manages to identify text properly but its the inbuilt dictionary which generally spoils the show by trying to make some sense of things it has read.

So, are you fully protected from automated scripts filling your forms??? For now, almost yes but beware...software grows real fast and people will come out with some newer techiniques.

To verify this, I used an image2text software and found that more than 70% of times the software actually manages to identify text properly but its the inbuilt dictionary which generally spoils the show by trying to make some sense of things it has read.

I was not aware of that. Interesting information. thanks!

bravo..

didint know abt this thing in such a deep!!! just took it as an identification code or may be a sort of distinction made in the users by the mail provider....

awesum job by biggie , jerry and pradypop

well now one more question on this... these codes when we see are in a distorted form.. sum twisted and stuff... so it is becoz of that inbuilt dictionary that pradypop quoted above or does it has sum different reason????

do enlighten me on dis as well....

What i meant by varying fonts and sizes meant that the letters could be twisted, elongated...So that only humans intrepet the text. Intended to decieve image2text softwares.

How image verification work?

i'll reply to this first...
when u request a captcha' page, a string is generated from a set of characters and converted into image using some "text to image" software. also this string is assigned to a session variable. when the page is submitted this string is compared with the one submittted by the user.
both must match AND this much be the first thing to do on reading the form data.

now comes why???
well quite already explained.. the only motive is to differentiate between human and computers. there are other captcha like one which put GK questions to its users, another that ask the model of the car shown in the image etc. but remember: captchas give a lot of false positives.
(this CE site captcha also bugged me a lot, when u have a distorted character u cant say S is s or otherwise;
Moral: chose the captcha characters carefully)

Jcaptcha and SimpleCaptcha are common java implementations... then there are ASP, PHP etc. based solutions as well...

hope information is useful

Old Thread !! But interesting information

Is there any Captcha API available for java to implement this in web application[ it should be more secure]?

I was aware of captcha ... but this was really informatic .... thnks big k and pradya

Can captcha be fooled? I've seen spambots register at another php forum where they got a captcha system implemented at registration.

CAPTCHA has been broken already. reCAPTCHA is yet to be broken (to the best of my knowledge)

Then what's the sophistication in reCAPTCHA that's not in CAPTCHA?
How's the former more secure than the latter??