[Bug-ocrad] Re: Problems recognizing character integrity

[Antonio Diaz Diaz]

Hello Uwe,

Uwe Dippel wrote:
> My excuses for running into another problem here:

Please, don't excuse yourself. Bug reports are important for the 
improvement of a program, and your reports are among the most useful I 
receive. Thanks.


> To me, the 'm' of 'my' is very much a single, singular character.
> I'm running 0.17 with the latest patch for 'J'.
> Unfortunately, this 'm' is recognized as 'rn', without ambiguity:

I hope the following patch fixes the problem.

--- character_r11.cc~   2007-06-29 14:29:13.000000000 +0200
+++ character_r11.cc    2007-07-17 12:54:32.000000000 +0200
@@ -201,6 +201,8 @@
         f.lp.minima() == 2 && f.rp.minima() == 2 )
       { add_guess( '*', 0 ); return; }
     if( b.id( b.vcenter(), b.hcenter() ) == 0 &&
+        b.id( b.vcenter() - 1, b.hcenter() ) == 0 &&
+        b.id( b.vcenter() + 1, b.hcenter() ) == 0 &&
         b.seek_left( b.vcenter(), b.hcenter() ) <= b.hpos( 25 ) )
       {                                        // Found merged 'rn'
       int row = b.vpos( 95 );


Regards,
Antonio.