First, it seems that the orthography colourblindness is rather unusual. I am not sure I can strongly defend it, but it’s the spelling I’ve been using. (Canadian OED lists colour-blindness, which I might be willing to write if I also used locutions like cocoa-nuts.) Anyway, if you’re trying to Google for the concept in question, you end up having to search for colour blindness, color blindness, colorblindness, colour deficiency, and colour deficiency, too.
Now, then: Ken Wakita and Kenta Shimamura have written a paper, “SmartColor: Disambiguation Framework for the Colorblind,” which sets out some algorithms to automatically convert colour in images (mostly charts and illustrations) into analogues that colourblind people can distinguish.
We already have several such systems, like Vischeck and aDesigner (both cited by the authors), but their algorithm allows a creator to specify that only colours above a certain threshold of importance should be changed. In theory, that means only charts and graphs could be adapted (examples), while photographs could be left alone.
Interestingly, the authors claim that anomalous tritanopia (a kind of weak blue deficiency) exists, even though everybody I talked to told me it doesn’t. They do a cute experiment in hacking the Ishihara colourblindness test so you can actually pass it even if you have a colour deficiency.
Now, how does this apply to Web sites? I’m not sure.
SmartColor assists the colourblind by grasping the author’s intention behind his/her colour usage and attempts to find a colour repainting that most effectively [imparts] the intention to the colourblind person. In this sense it is a semantic approach. […]
[A] promising application domain is production of stylesheets such as CSS for HTML and XSL for XML. The SmartColor algorithm takes a Web page and produces a colouring method as a stylesheet for… each class of colourblindness.
Essentially, you could have a user CSS file that redesigns (“repaints”) Web sites on the fly. Except that the CSS would not be all you’d need; you’d need an extension (Firefox, anyone?) that acts as a proxy server and custom-creates a new CSS for each page.
In any event, that would be useful only for encoded text, not for text inside graphics or for any kind of graphics other than SVG, which nobody uses.
Moreover, and this is not the authors’ fault, it seems impossible, after years of trying, to get the Web Accessibility Initiative clear on what colour deficiency actually is and what it means for Web usage. In many years of checking, I have found only a few dozen pages (not even sites, just pages) that could even possibly present a problem to a colour-deficient person. If you follow WCAG to the letter, you destroy much of the design of your Web page that is helpful to all visual users except low-vision people, and you also don’t necessarily or even usually help colourblind people.