1SHAPECLUSTERING(1)                                          SHAPECLUSTERING(1)
2
3
4

NAME

6       shapeclustering - shape clustering training for Tesseract
7

SYNOPSIS

9       shapeclustering -D output_dir -U unicharset -O mfunicharset -F
10       font_props -X xheights FILE...
11

DESCRIPTION

13       shapeclustering(1) takes extracted feature .tr files (generated by
14       tesseract(1) run in a special mode from box files) and produces a file
15       shapetable and an enhanced unicharset. This program is still
16       experimental, and is not required (yet) for training Tesseract.
17

OPTIONS

19       -U FILE
20           The unicharset generated by unicharset_extractor(1).
21
22       -D dir
23           Directory to write output files to.
24
25       -F font_properties_file
26           (Input) font properties file, where each line is of the following
27           form, where each field other than the font name is 0 or 1:
28
29               'font_name' 'italic' 'bold' 'fixed_pitch' 'serif' 'fraktur'
30
31       -X xheights_file
32           (Input) x heights file, each line is of the following form, where
33           xheight is calculated as the pixel x height of a character drawn at
34           32pt on 300 dpi. [ That is, if base x height + ascenders +
35           descenders = 133, how much is x height? ]
36
37               'font_name' 'xheight'
38
39       -O FILE
40           The output unicharset that will be given to combine_tessdata(1).
41

SEE ALSO

43       tesseract(1), cntraining(1), unicharset_extractor(1),
44       combine_tessdata(1), unicharset(5)
45
46       https://tesseract-ocr.github.io/tessdoc/Training-Tesseract.html
47

COPYING

49       Copyright (C) Google, 2011 Licensed under the Apache License, Version
50       2.0
51

AUTHOR

53       The Tesseract OCR engine was written by Ray Smith and his research
54       groups at Hewlett Packard (1985-1995) and Google (2006-present).
55
56
57
58                                  03/20/2023                SHAPECLUSTERING(1)