|
I will add to that too and also give the same opinion. PDFs are usually designed to be printed and are made in programs like InDesign, Quark, and Acrobat which pretty much work as WYSIWYG (what you see is what you get) editors.
Most of the text is done in individual boxes, one for the heading, one for each paragraph, column, etc. And they're layered, so you're just hoping that the writer did add them one after another, which is never the case. This becomes apparent when you're making selections and something else is being highlighted.
Stick any PDF document into Adobe's Acrobat editor, and you literally see how awful the setup is.
I would think OCR would work better with a flattened image, as long as it was 300dpi or more.
|