Changelog of the OCR Toolkit for Gamera
=======================================

Version 1.0.6, Feb 13, 2012
---------------------------

 - extrachars file can contain arbitrary unicode texts

 - heuristic rule for disambiguating roman letters somewhat improved


Version 1.0.5, Sep 19, 2011
---------------------------

 - FSF address corrected


Version 1.0.4, Jul 27, 2010
---------------------------

 - grouping of quotation marks somewhat corrected
   (should this be removed some day?)

 - more heuristic rules for disambiguation of roman characters

 - heuristic rules are now choosable in ocr4gamera.py with option "-R"
   to switch them off, use "-R none"


Version 1.0.3, Jun 15, 2010
---------------------------

 - Option "-o" now works again


Version 1.0.2, Jun 08, 2010
---------------------------

 - changes in recognition script "ocr4gamera.py":
    o renamed from "ocr4gamera" to "ocr4gamera.py" (for Windows compatibility)
    o changes in dictionary correction
       + language choosable with option "-L"
       + edit distance threshold choosable with option "-e"
       + when aspell not found, ispell is tried
    o additional translation table class_name -> character
      can be provided in a CSV file with option "-c"
    o option "-i" removed; replaced with verbosity level in option "-v"
    o option "-a" for grouping of broken characters now works

 - new option extra_chars_dict in functions
   return_char() and textline_to_string()

 - new plugin bbox_merging_mcmillan in analogy to roman_text

 - plugin bbox_seg removed (now in the Gamera core)

 - replaced the non standard module "pexpect" with "popen"


Version 1.0.1, Feb 25, 2010
---------------------------

 - 'ocr4gamera -a' now uses the provided training file

 - ocr4gamera no behaves properly when no image file given

 - GPL copyright information added to all source file;
   documentation licensed under CC-BY


Version 1.0.0, Oct 16, 2009
---------------------------

 - first creation
