There are just some observations here on our side after some tests of
today with respect to width of essentially the same font.
One sample is a built-in font, using lpr {filename}; while the other is
the same file in KWrite, printed with CUPS. KWrite, for one reason or
another beyond our intention, squeezes the characters very close, it
looks like highly over-kerned.
While we surely don't expect the overdone kerning to recognise properly,
we are still astonished that the wide character distance results in
flawless recognition, while the overkerned sample misses out one clearly
existing letter, an 'i', with its characteristic dot. Also, a very
obvious third vertical stroke of a 'm' is definitively discarded.
We have scanned both another time, at a rotation of 180°, and found
identically the same result.
Again, we are happy with what we get, though would like to raise our
concern about the non-recognition of two unambiguous features; the
existence of a letter 'i' and three strokes in an 'm' that are
recognized reproducibly as 'n'.
Uwe
_______________________________________________
Bug-ocrad mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-ocrad
