Adaptive OCR for books

An adaptive optical character recognition (OCR) system for a book having at least one page with known and unknown fonts comprising:

• the entire book, wherein the entire book is scanned and OCR of the entire book is performed;
• a verification process applied for the known fonts, wherein the unknown fonts are clustered by shape;
• at least one sure word separated out from at least one error words,
• wherein the sure words and error words are sent to an OCR training,
• wherein outliers are determined,
• wherein the unknown fonts are manually classified,
• wherein the manually classified fonts are sent to OCR training and the manually classified fonts are processed through the verification for known fonts and the sure words are separated out from error words;
• manually classified results are processed to a quality evaluator;
• wherein the sure words are placed in a depository if they meet a predetermined quality criteria and sure words, error words and manually classified fonts are processed as required using an iterative process until all fonts meet the predetermined quality criteria.