Almost everyone has seen many captcha images on quite all web forms.
The security would have to be granted by the fact that, while for a human it pretty easy to recognoise a sequence of symbols drawn, that task could be hard for a computer that processes the image with the intent of figuring out the required sequence.
That gap dramatically grows as far as some cognitive factors are employed within the process. This imlpies not only the recognition of the sequence but even some kind of reasoning on that. An example could be something like listing only a ‘subcharset’ in the sequence. That ensure a good level of anti-spam form security.
What about the nature of the image? I mean, how much can you decide about the aspect of the image (colors, noises, effects, charsets, etc. …)?
I made a class that allows to decide almost every aspect of the resulting image with minimal configuration (through array).
Get a look at theese samples.
The class “ucaptcha” is available here.