paperkey // recommended OCR font ?

[James Cloos]

>>>>> "v" == vedaal  <vedaal at nym.hush.com> writes:

v> If so, what is the recommended font and size to be used for accurate OCR ?

v> OCR-A, OCR-B,  Ordinary Courier 10,  other ?

I've tried it once.  I used ocr-a since it was available and seemed
likely to be easiest to scan.  And, although some might disagree, I
also find it easy to read.

I just gave it a try w/o paper, using mpage to generate postscript,
OCRA for the font, gs to render the ps to pbm, and gocr to extract
the text.  I needed to run >>tr \;_ ::<< on the extracted text, but
with that paperkey was able to generate a new secring.

So be aware that you may need to massage the ocr'ed data to recover
paperkey's format, but it should mostly work.

Using OCRB or Courier, gocr wasn't able to recover the text well enough.

Tesseract did better with Courier-Bold, but needed >>tr Ol 01<<.

But tesseract *badly* fails to grok OCRA!

I suspect that actual printing and scanning won't be *too* much worse.

-JimC
-- 
James Cloos <cloos at jhcloos.com>         OpenPGP: 1024D/ED7DAEA6