On 2022-04-27 9:14 a.m., Paul Koning via cctalk wrote:
...
Re OCR-B: the difference between zero and O in that font is small enough that
contemporary OCR could not reliably tell the two apart. This is documented in detail in
"Travels in Computerland" by Ben R. Schneider, a book about his project to
digitize a multi-volume printed document in the early 1970s. It involved having it typed
(in Hong Kong I think) using OCR-B type balls, and when they ran into the OCR issue it
was worked around by modifying the type balls to give one of the two characters a cut in
the left side, making it like a reversed C. OCR sure has come a long way since then.
Since OCR-B has been mentioned people might be interested in this paper
on its development:
http://www.telegraphics.com.au/doc/scarrott.html
--Toby
Yes, OCR-A is extremely ugly; Schneider actually considered it before
dismissing it, on the grounds its letter forms are so bad that proofreaders
trying to check the as-typed material would have a hard time dealing with it
and quality would suffer as a result.
paul
