Content Based Image Analysis

[learning_brain]

Hi

I've been working on a php class now for about a fortnight.

The class breaks down the image into critical components and patterns and compares it to a database of other image "fingerprints", before scoring it.

The result is that the class will return the closest description match - meaning that it can tell (% chance) what an image is before displaying it

demo

It looks messy at the mo because I am still going through beta de-bugging and increasing the image comparison database, but you get the gist...

The class call will be as simple as...

$analysedImage = new ImageAnalysis();
$analysedImage->doAnalysis($source_file);

where $source_file is the URL of an image.


MY QUESTION IS.....

Do I

a) tidy up and sell the class as it stands...
b) tidy up and concentrate on a nudity risk so that it can return a warning... then sell as a class
c) tidy up and then produce a "safe-view" premium site that acts a bit like diggit (framed) only with all img tags on sites viewed replaced by this class...

Any comments are greatly appreciated.

Thanks

Richard