Some years back I scanned a Swedish-Latin dictionary (1875), 960 pages in 2 volumes, with the Swedish headwords printed in blackletter and the Latin translations in plain Roman letters, http://runeberg.org/swelatin/
Johan Winge decided to OCR and proofread it, and for this purpose he trained Tesseract, the free software OCR engine, to read blackletter.
He also developed a program called Terese for proofreading the OCR text, http://terese.sourceforge.net/
It looks very promising from the description, but I haven't tried it yet. The screenshots in the documentation shows pages from the Swedish-Latin dictionary.