ocre - ocr e..... (to fill)
Options (ephemeral): [-h] [-d ndeb] [-l
language] [-p nproc] [-y n]
[-C char] [-D ]
recognises characters from the images in the files nfImage
... . The input files must be in pgm
(raw) format (256 gray levels), in
(raw) format, or in jpg
format. At the standard output we get
the recognised text.
Some of the present options have been introduced for help in debugging and
building a font data base employed to teach the system. There is no commitment
for supporting them.
- -d ndeb
- Writes debugging messages at the standard output.
- Writes a help message with options.
- -l language
- Expects language in the document. Opens a window with specific
characters of the language. Adapts the output to the language. (for
instance ascii 8859-2 if language is 'pol' (polish)).
- -p nproc
- Process to do with the images. If nproc is 3, shows the segmented
characters. If nproc is 5, writes the characters from the image,
each one in a file, under ./arm1 .
- -y n
- Writes page number: with tags (1), first line (2), only (3).
- -C char
- writes a enlarged average version of character char.
- uses aspell to fill some words.
- Directory with character decision trees.
- decision trees index.
- decision trees.
Luis Cearra. http://lem.eui.upm.es/luisjc.html
Does not recognise well large characters, very little ones, two columns, ..., I
am working in it . This is a small model. :-) .