wallix / ppocr

Basic OCR.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

LGPLv2.1

Learning (./learning)

GĂ©nerating glyphs/* and datas/*:

$ ./learning "$bindir"/write_glyphs "$bindir"/normalize_glyphs "$bindir"/write_datas

The executables are compiled with bjam. bindir depends on the compiler used.

$ bjam or $ bjam write_glyphs normalize_glyphs write_datas

Note: an error ("error: bad pnm type; expected P3 or P6, get P4!") is always displayed. It comes from Milena sources (include / mln), which displays an error when the file can not be read. As ppocr trying 2 formats (PBM and PNM), if the first reading fails, the error appears.

resources/fonts/{latin,cyrillic}/

./letters and ./generated/

letters contains glyphs of generated//.pnm (generated by font2png (font2png.cpp) then converted to pnm with pngtopnm (imagemagick))

resources/fonts/generated_extract_fail/

Images generated incorrectly (font with hyphens). The directory is ignored.

./mano/

Images belonging to the old ocr (OCRv1, see redemption project).

resources/dict_*.trie

Generated from resources/dict_*.txt by make_trie_dictionary

resources/glyphs/*

Generated from image files (font/*) by write_glyphs_bin and normalizer_glyphs_bin

resources/datas/*

Generated from data glyphs (glyphs/*) by write_datas

About

Basic OCR.

License:GNU Lesser General Public License v2.1


Languages

Language:C++ 99.7%Language:Shell 0.3%