OCRopus - open source document analysis and OCR system (www.ocropus.org)


--------------------------------------------------------------------------------
current
--------------------------------------------------------------------------------
    * jam is replaced by scons
    * automake support
    * image understanding related code is now in separate iulib
    * OCRopus requires installed iulib (http://code.google.com/p/iulib)


--------------------------------------------------------------------------------
Version 0.2 (2008-05-30) "Alpha 2"
--------------------------------------------------------------------------------

   - fixed dependency problems in jam
   - graphical logging and debugging facility
   - new beam search decoder
   - make use of OpenFST and Tesseract optional
   - implement hOCR output in lua
   - reduced memory usage of layout analysis
   - xy-cut layout analysis
   - Voronoi-based layout analysis (donated by K. Kise)
   - Otsu binarization (in addition to our own fast Sauvola)
   - script-based automatic EM training
   - ocropus and libraries callable directly from Python/NumPy
   - replacement of ocrocmd by customizable Lua-scripts
   - various bug fixes 

--------------------------------------------------------------------------------
Version 0.1.1 (2007-12-14)
--------------------------------------------------------------------------------
This is the first maintenance release of OCRopus Alpha, which introduces closer 
cooperation with Tesseract to improve speed and accuracy. It also fixes several 
portability issues.

New Features:
    * compatibility with Mac OS X and Windows
    * ocrocmd uses block segmenter with Tesseract
    * Lua interpreter and tolua++ integrated into the package

Fixes:
    * hOCR output is now valid (and useful) XHTML
    * some portability fixes
    * cleaned up deprecated interfaces


--------------------------------------------------------------------------------
Version 0.1.0 (2007-10-22)
--------------------------------------------------------------------------------
The first packaged release aims to stabilize the interfaces of the involved
components and provides a first version of the scripting functionality
for OCRopus based on Lua. It also includes some more preprocessing functionality
for document images and a MLP based character classifier.

New Features:
    * Lua-based scripting of ocropus (ocroscript)
    * unit and functional testing via ocroscript
    * text/image segmentation in document images
    * document image cleanup
    * document image deskewing
    * MLP-based character recognition
    * OpenFST-based statistical language modeling
    * fast binary morphology
    * use of Tesseract 2.x instead of 1.x
    * alignment and training data generation from transcribed ground truth

Fixes:
    * better code organization through namespaces, include file simplifications
    * many refactorings of core components for better maintainability

--------------------------------------------------------------------------------
Technology Preview Announcement  (2007-04-09)
--------------------------------------------------------------------------------
The Ocropus svn was opened to the public in April 2007 in order to give a first
preview of the project. The Technology Preview internally combined Tesseract 1.x
with some Layout Analysis algorithms.In this stage, a basic system was already
working but the whole architecture was still changing as well as the interfaces
for the respective modules / components.
