Recognizing known text (generating searchable PDF)

[Erik Jensen]

I'm trying to generate a searchable and copyable PDF from a series of images. Using Tesseract works pretty well, but still results in a number of errors on each page. However, I already have a copy of the text that appears on each page, so all I really need is to find the location of each of the known glyphs on the page so I can put the overlay text in the correct location. Is there a way to use the known text to guide Tesseract's recognition to accomplish this?

Thanks.