> Then the image
>has to be segmented into words, and words into letters; each letter is
>then recognized, and usually a confidence value is attached to each
>letter.
Apple was doing work on handwriting recognition and had phenomenally
high recognition rates by looking at words rather than letters. Of
course this is not language independent and requires dictionaries -
but these can be built up by the software (training).
An ideal OCR would use the "lower level" results from the letter
recognition to influenece the "high level" word recognition and
vice versa.
I'm not familiar with this Neural Net speak, but I think this
sort of thing is possible and should help to achieve much higher
accuracy levels.
Cheers,
Colin
---------------------------------------------------------------
Colin 't Hart, Systems Analyst Email: cthart@gbs.com.au
Geographic Business Systems Pty Ltd
The Garden Office Park Phone: +61 8 9443 8822
345 Harborne St (PO Box 761) Fax: +61 8 9242 4412
Herdsman WA 6017
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com