[OpenCV-EMGU 記錄] Tesseract training

1. Create target *.tif which is set 2 colors: black and white).

2. Make Box file.
tesseract eng.segment7.exp0.tif eng.segment7.exp0 batch.nochop makebox

3. Make Tr file.
tesseract eng.segment7.exp0.tif eng.segment7.exp0 box.train

4. Make unicharset file.
unicharset_extractor eng.segment7.exp0.box

5. Create a file named "font_properties" without txt.

content: segment7 0 0 0 0 0

6. Make inttemp/pffmtable/shapetable files.
mftraining -F font_properties -U unicharset eng.segment7.exp0.tr

7. Make normproto file.
cntraining.exe eng.segment7.exp0.tr

8. Rename normproto/inttemp/pffmtable/shapetable/unicharset files with prefix "eng.".

9. Make traineddata file.
combine_tessdata eng


//shapeclustering -F font_properties -U unicharset eng.segment7.exp0.tr