Say you write

THE CAT

but it looks more like

TAE CHT

because something (perhaps a cat) distracted you while you were writing.

The OCR today would give: Tae cht
with a combo box allowing you to chose corrections. After the correction the text will look so: The cat

My proposal. Why not have letters that can be something between an A and an H, so that if the software cannot decide, it will show that letter, and leave it to the reader to decide what the letter is.

Since all computers will have these new hybrid letters, there's no need to fix anything during copy and paste.

If you still want to fix the resulting text, there's nothing easier. Find all the hybrids, and simply chose between one of the two or three choices.
-- pashute, Oct 25 2011

Do they not use context-based algorithms (thinking T9 or Apple's iType thingy) in order to help cleanse OCR results?

The trouble with using hybrids, is that if you're not careful, everything starts looking like a hybrid of something or another.
-- zen_tom, Oct 25 2011


This just ends up presenting the page of text as a JPG image.
-- pocmloc, Oct 25 2011


Drat. I wanted this to be a font which reads normally to human eyes, but which yields obscenities or insults when OCR'd.
-- mouseposture, Oct 26 2011



random, halfbakery