[Bug-ocrad] Re: Error recognising capital 'J'

[Antonio Diaz Diaz]

Hello Uwe,

Ocrad 0.17 has changed. It now recognizes it as a capital 'I'. Most 
probably ocrad 0.18 will recognize it correctly.

By the way, if correctly converted JasJaspal.pbm.bz2 is only 1136 bytes 
long, not huge at all.


Regards,
Antonio.


Uwe Dippel <address@hidden> wrote:
> We found an error at recogition of 'J', at least in 0.16.
> Even at a rather clear height, it always ends at '7', and even 
> unambiguously:
>
> # Ocr Results File. Created by GNU Ocrad version 0.16
> source file JasJaspal.pnm
> total text blocks 1
> text block 1 0 0 705 129
> lines 1
> line 1 chars 35 height 22
>  95  46 13 26; 1, '7'0
> 111  54 16 18; 1, 'a'0
> 128  54 13 18; 1, 's'0
> 142  54 18 26; 1, 'p'0
> 163  54 16 18; 1, 'a'0
> 180  45  8 27; 2, '1'1, 'l'0
> 187  45 11 35; 1, 'j'0
> 202  54 16 19; 1, 'e'0
> 220  54 15 19; 1, 'e'0
> 237  49 10 24; 1, 't'0
>
> I have added the .tif for your information (pnm is so huge ...)
> What I did:
> tifftopnm JasJaspal.tif > JasJaspal.pnm
> ocrad -x JasJaspal.orf JasJaspal.pnm
>
> Thanks for looking into it - or has 0.17 been changed w.r.t. this ?