Databases Reference
In-Depth Information
Symbol Region Decoding
The symbol region decoding procedure is a dictionary-based decoding procedure. The symbol
region segment is decoded with the help of a symbol dictionary contained in the symbol
dictionary segment. The data in the symbol region segment contains the location where
a symbol is to be placed, as well as the index to an entry in the symbol dictionary. The
symbol dictionary consists of a set of bitmaps and is decoded using the generic decoding
procedures. Note that because JBIG2 allows for lossy compression, the symbols do not have
to exactly match the symbols in the original document. This feature can significantly increase
compression performance when the original document contains noise that may preclude exact
matches with the symbols in the dictionary.
Halftone Region Decoding
The halftone region decoding procedure is also a dictionary-based decoding procedure. The
halftone region segment is decoded with the help of a halftone dictionary contained in the
halftone dictionary segment. The halftone dictionary segment is decoded using the generic
decoding procedures. The data in the halftone region segment consists of the location of
the halftone region and indices to the halftone dictionary. The dictionary is a set of fixed-
size halftone patterns. As in the case of the symbol region, if lossy compression is allowed,
the halftone patterns do not have to exactly match the patterns in the original document. By al-
lowing for nonexact matches, the dictionary can be kept small, resulting in higher compression.
7.8 MRC-T.44
With the rapid advance of technology for document production, documents have changed
in appearance. Where a document used to be a set of black and white printed pages, now
documents contain multicolored text as well as color images. To deal with this new type
of document, the ITU-T developed recommendation T.44 for Mixed Raster Content (MRC).
This recommendation takes the approach of separating the document into elements that can
be compressed using available techniques. Thus, it is more an approach of partitioning a
document image than a compression technique. The compression strategies employed here
are borrowed from previous standards such as JPEG (T.81), JBIG (T.82), and even T.6.
The T.44 recommendation divides a page into slices where the width of the slice is equal
to the width of the entire page. The height of the slice is variable. In the base mode, each
slice is represented by three layers: a background layer, a foreground layer, and a mask layer.
These layers are used to effectively represent three basic data types: color images (which
may be continuous tone or color mapped), bi-level data, and multilevel (multicolor) data.
The multilevel image data is put in the background layer, and the mask and foreground lay-
ers are used to represent the bi-level and multilevel nonimage data. To work through the
various definitions, let us use the document shown in Figure
7.16
as an example. We have di-
vided the document into two slices. The top slice contains the picture of the cake and two lines
