How can I extract the Text from a GIF Image? Please help!

[Guest]

Hi!

I'm trying to code a testing application for my company's website which will
try out all combinations of option buttons, list-box selections etc to make
sure everything works out as expected.

Unfortunately I've hit a brick wall when faced with the GIF "random security
code generator" image. Could someone please detail how to decrpyt the text in
the gif? Any API functions or similar that can be used to assist in this goal?

The URL for the GIF pwd generator is available here (just temporarily until
I can get past it with your help!):

<www.nrma.com.au/PremiumEstimates/pepasswordgifgenerator?>

Many thanks in advance everyone! I really appreciate any light you can shed
on this!

 

[NickHK]

Jas,
Isn't that the whole point of these; they virtually impossible to decypher
by code/robots.

NickHK

 

[Guest]

The "text" in a gif is not in text (string) form, it is a picture so there is
no real way to extract it except OCR - but these security images are designed
to be resistant to attempts to read/convert them, and as Nick said the reason
for that should be obvious. Frankly, if you or anyone were successful in
finding a way to read the text with an automated process, you should tell
your company that their security is not adequate. So the best result you
should hope for is that your attempt will fail!

 

[Peter T]

I would not be at all surprised if this is not already possible, if not yet
then soon.

Search terms: Cognitive-recognition, Pattern-recognition, ai,
neural-networks

Regards,
Peter T

 

[Guest]

OCR is definitely possible, and widely used - but the specific patterns used
to distort the letters to assure a human is reading the page (and typing in
the text) are designed specifically to be unreadable by automated means. If
not it would be very poor security - and that is my point.

 

[Peter T]

I wasn't trying to be contrary, but OCR is not quite what I had in mind.
This illustration was given to me by someone working on this topic:

Imagine a crate of fish, a mixture of species packed un-arranged, with
crushed ice partially covering them. A fisherman will easily identify each
fish. But how to image process a picture of that crate, and supply a list of
each partially visible fish.

When this told to me about two years ago it was a challenge, being
intensively worked on. I'll try and find out if it has been cracked!

Regards,
Peter T

 

[Tim Williams]

This makes interesting reading (see "Defeating CAPTCHAS").

http://nilesh.org/weblog/categories/security/

Seems nothing is beyond the will of the determined...

Tim