I think I mentioned it before, but you could go further with the Go computer vision thing—Ken Thompson considerably improved OCR results for scanning chess books by adding in domain-specific knowledge (about chess): http://doc.cat-v.org/bell_labs/reading_chess/
I think I mentioned it before, but you could go further with the Go computer vision thing—Ken Thompson considerably improved OCR results for scanning chess books by adding in domain-specific knowledge (about chess): http://doc.cat-v.org/bell_labs/reading_chess/