summarylogtreecommitdiffstats
path: root/jbig2enc.1
blob: c7ccdff75cad7d6405f9672b7460f459b421a346 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
.\" Automatically generated by Pandoc 2.17.1.1
.\"
.\" Define V font for inline verbatim, using C font in formats
.\" that render this, and otherwise B font.
.ie "\f[CB]x\f[]"x" \{\
. ftr V B
. ftr VI BI
. ftr VB B
. ftr VBI BI
.\}
.el \{\
. ftr V CR
. ftr VI CI
. ftr VB CB
. ftr VBI CBI
.\}
.TH "JBIG2ENC" "1" "" "" ""
.hy
.SH jbig2enc: Documentation
.PP
Adam Langley <agl\[at]imperialviolet.org>
.SS What is JBIG2
.PP
JBIG2 is an image compression standard from the same people who brought
you the JPEG format.
It compresses 1bpp (black and white) images only.
These images can consist of \f[I]only\f[R] black and while, there are no
shades of gray - that would be a grayscale image.
Any \[dq]gray\[dq] areas must, therefore be simulated using black dots
in a pattern called halftoning (http://en.wikipedia.org/wiki/Halftone).
.PP
The JBIG2 standard has several major areas:
.IP \[bu] 2
Generic region coding
.IP \[bu] 2
Symbol encoding (and text regions)
.IP \[bu] 2
Refinement
.IP \[bu] 2
Halftoning
.PP
There are two major compression technologies which JBIG2 builds on:
arithmetic encoding (http://en.wikipedia.org/wiki/Arithmetic_coding) and
Huffman encoding (http://en.wikipedia.org/wiki/Huffman_coding).
You can choose between them and use both in the same JBIG2 file, but
this is rare.
Arithmetic encoding is slower, but compresses better.
Huffman encoding was included in the standard because one of the
(intended) users of JBIG2 were fax machines and they might not have the
processing power for arithmetic coding.
.PP
jbig2enc \f[I]only\f[R] supports arithmetic encoding
.SS Generic region coding
.PP
Generic region coding is used to compress bitmaps.
It is progressive and uses a context around the current pixel to be
decoded to estimate the probability that the pixel will be black.
If the probability is 50% it uses a single bit to encode that pixel.
If the probability is 99% then it takes less than a bit to encode a
black pixel, but more than a bit to encode a white one.
.PP
The context can only refer to pixels above and to the left of the
current pixel, because the decoder doesn\[aq]t know the values of any of
the other pixels yet (pixels are decoded left-to-right, top-to-bottom).
Based on the values of these pixels it estimates a probability and
updates it\[aq]s estimation for that context based on the actual pixel
found.
All contexts start off with a 50% chance of being black.
.PP
You can encode whole pages with this and you will end up with a perfect
reconstruction of the page.
However, we can do better...
.SS Symbol encoding
.PP
Most input images to JBIG2 encoders are scanned text.
These have many repeating symbols (letters).
The idea of symbol encoding is to encode what a letter \[lq]a\[rq] looks
like and, for all the \[lq]a\[rq]s on the page, just give their
locations.
(This is lossy encoding)
.PP
Unfortunately, all scanned images have noise in them: no two
\[lq]a\[rq]s will look quite the same so we have to group all the
symbols on a page into groups.
Hopefully each member of a given group will be the same letter,
otherwise we might place the wrong letter on the page!
These, very surprising, errors are called cootoots.
.PP
However, assuming that we group the symbols correctly, we can get great
compression this way.
Remember that the stricter the classifier, the more symbol groups
(classes) will be generated, leading to bigger files.
But, also, there is a lower risk of cootoots (misclassification).
.PP
This is great, but we can do better...
.SS Symbol retention
.PP
Symbol retention is the process of compressing multi-page documents by
extracting the symbols from all the pages at once and classifing them
all together.
Thus we only have to encoding a single letter \[lq]a\[rq] for the whole
document (in an ideal world).
.PP
This is obviously slower, but generates smaller files (about half the
size on average, with a decent number of similar typeset pages).
.PP
One downside you should be aware of: If you are generating JBIG2 streams
for inclusion to a linearised PDF file, the PDF reader has to download
all the symbols before it can display the first page.
There is solution to this involing multiple dictionaries and symbol
importing, but that\[aq]s not currently supported by jbig2enc.
.SS Refinement
.PP
Symbol encoding is lossy because of noise, which is classified away and
also because the symbol classifier is imperfect.
Refinement allows us, when placing a symbol on the page, to encode the
difference between the actual symbol at that location, and what the
classifer told us was \[lq]close enough\[rq].
We can choose to do this for each symbol on the page, so we don\[aq]t
have to refine when we are only a couple of pixel off.
If we refine whenever we a wrong pixel, we have lossless encoding using
symbols.
.SS Halftoning
.PP
jbig2enc doesn\[aq]t support this at all - so I will only mention this
quickly.
The JBIG2 standard supports the efficient encoding of halftoning by
building a dictionary of halftone blocks (like the dictionaries of
symbols which we build for text pages).
The lack of support for halftones in G4 (the old fax standard) was a
major weakness.
.SS Some numbers
.PP
My sample is a set of 90 pages scanning pages from the middle of a
recent book.
The scanned images are 300dpi grayscale and they are being upsampled to
600dpi 1-bpp for encoding.
.IP \[bu] 2
Generic encoding each page: 3435177 bytes
.IP \[bu] 2
Symbol encoding each page (default classifier settings): 1075185 bytes
.IP \[bu] 2
Symbol encoding with refinement for more than 10 incorrect pixels:
3382605 bytes
.SS Command line options
.PP
jbig2enc comes with a handy command line tool for encoding images.
.TP
\f[B]-d\f[R] | \f[B]--duplicate-line-removal\f[R]
When encoding generic regions each scan line can be tagged to indicate
that it\[aq]s the same as the last scanline - and encoding that scanline
is skipped.
This drastically reduces the encoding time (by a factor of about 2 on
some images) although it doesn\[aq]t typically save any bytes.
This is an option because some versions of jbig2dec (an open source
decoding library) cannot handle this.
.TP
\f[B]-p\f[R] | \f[B]--pdf\f[R]
The PDF spec includes support for JBIG2
(Syntax\[->]Filters\[->]JBIG2Decode in the PDF references for versions
1.4 and above).
However, PDF requires a slightly different format for JBIG2 streams: no
file/page headers or trailers and all pages are numbered 1.
In symbol mode the output is to a series of files: symboltable and
page-\f[I]n\f[R] (numbered from 0)
.TP
\f[B]-s\f[R] | \f[B]--symbol-mode\f[R]
use symbol encoding.
Turn on for scanned text pages.
.TP
\f[B]-t\f[R] <threshold>
sets the fraction of pixels which have to match in order for two symbols
to be classed the same.
This isn\[aq]t strictly true, as there are other tests as well, but
increasing this will generally increase the number of symbol classes.
.TP
\f[B]-T\f[R] <threshold>
sets the black threshold (0-255).
Any gray value darker than this is considered black.
Anything lighter is considered white.
.TP
\f[B]-r\f[R] | \f[B]--refine\f[R] <tolerance>
(requires \f[B]-s\f[R]) turn on refinement for symbols with more than
tolerance incorrect pixels.
(10 is a good value for 300dpi, try 40 for 600dpi).
Note: this is known to crash Adobe products.
.TP
\f[B]-O\f[R] <outfile>
dump a PNG of the 1 bpp image before encoding.
Can be used to test loss.
.TP
\f[B]-2\f[R] or \f[B]-4\f[R]
upscale either two or four times before converting to black and white.
.TP
\f[B]-S\f[R]
Segment an image into text and non-text regions.
This isn\[aq]t perfect, but running text through the symbol compressor
is terrible so it\[aq]s worth doing if your input has images in it (like
a magazine page).
You can also give the \f[B]--image-output\f[R] option to set a filename
to which the parts which were removed are written (PNG format).