Use Computer Vision to test how similar I I L and 1 are

[davelab6]

• Make a downstream test, note level, using a computer vision library like simplecv.org or similar, to test how similar capital i, lowercase l, capital L, and digit 1 look.

I'm not sure how to best express the similarity of the 4. I guess there if you average all 4 instances of a letterform, then measure the distance from the average to each instance, and the higher the distance, the more different it is?

And then if the distance is below a certain common amount, we could fail the test.

But until the test is run on 100s of fonts, we can't know what that limit would be...