Shortcomings and Alternatives

Free software OCR programs lack many of the tools of their proprietary counterparts. Neither Tesseract nor CuneiForm, for example, allow you to exclude image areas or work with files taken fresh from the scanner – a lack that adds an extra step to the process.

Probably the single most needed feature for both CuneiForm and Tesseract is the ability to recognize fonts. Although to the ordinary user, a font is a font, in practice, letter shapes can vary considerably between fonts. Recognizing fonts would also allow italic and cursive fonts to be converted to text with a much higher degree of accuracy.

However, the projects appear small, and OCR is a specialty that relatively few programmers are likely to work on unless they need it themselves. Consequently, users are likely to continue to get poor results at least part of the time.

When that happens, perhaps the best solution is to search for alternatives. If you can convert an image to PDF, you have several options for output to other formats, starting with scripts like pdf2ps, ps2ascii, and ps2text. Most of these scripts will extract text at least as accurately as Tesseract and CuneiForm, and some may even preserve layout better. OCR can be convenient when it works, but, in free software, it has yet to get the attention that it needs.