> On Fri, 25 Mar 2011, Matthew Toseland wrote: > >> Suggestions for text-based CAPTCHAs that computers cannot easily crack would >> be welcome.
Just a conceptual idea here--why not take CAPTCHA images and generate a Unicode representation? For each 'region' of the CAPTCHA image there would be thousands of possible characters that would suffice. If a random choice were made for such characters, such a process would produce a more secure text-based CAPTCHA as only the human eye would be able to easily understand the hidden message.
