CCITT T.6 2-Dimensional Compression
Standard
INTRODUCTION
In 1984, the CCITT developed a specification defining the facsimile coding
schemes to be used in the Group 4 black and white facismile. The standard, known
as CCITT Recommendation T.6, is designed for error-free digital facsimile
transmission mainly on public data networks. This specification has been widely
used in document imaging systems for compressing black and white images. It is
sometimes referred to as the CCITT Group 4 compression standard.
The CCITT T.6 facsimile coding standard consists of this basic facsimile
coding scheme and future optional facsimile coding schemes. It is assumed that
transmission errors are corrected by control procedures at a lower level.
Therefore, CCITT T.6 is a noiseless coding scheme.
Since the CCITT T.6 coding scheme is designed for noise free transmission, a
single bit error can cause corruption of the image. Compared to the CCITT T.4
coding scheme, this scheme is very sensitive to error, and is less robust.
HOW DOES CCITT T.6 WORK
The CCITT T.6 coding scheme is known as the Modified Modified Relative
element address designate code (MMR). The basic facsimile coding scheme uses
a two-dimensional line-by-line coding method, the position of each changing
picture element on the current coding line is coded with the respect to the
position of a corresponding reference element located on either the coding line
or the reference line. The reference line is immediately above the coding line.
After the coding line has been coded, it becomes the reference line for the next
coding line. The reference for the first line of a page is an imaginary white
line. This coding scheme is the same as the two-dimensional coding scheme of
Group 3 facsimile specified in CCITT T.4, except the first line is encoded
differently, and in the Group 3 two-dimensional coding scheme, every Kth line
(K=2 or 4) of image data must be encoded by the Group 3 one-dimensional coding
scheme.
The one-dimensional coding scheme of Group 3 facsimile makes use of only
horizontal dependencies between picture elements in the same scan line. The
advantage of the CCITT T.6 coding scheme over the one-dimensional coding scheme
of Group 3 facsimile is that it exploits both horizontal and vertical
dependencies, resulting in the greater compression efficiency. Since the CCITT
T.4 two-dimensional coding scheme requires that every Kth line of image data
must be encoded by the one-dimensional coding scheme, it is less efficient
compared to the CCITT T.6 coding scheme.
Group 4 encoding is much more complex. As mentioned earlier, each scanline is
compared with the previous one - in this way, vertical features in the source
image can be used to achieve better compression ratios. Rather than considering
the alternating white and black runs in each scanline, it considers the
positions of changing pixels, in other words the positions of the first pixel in
each run. Although these are effectively one and the same thing, Group 4 uses
the position of the changing pixel relative to that of the one on the previous
line as the coding scheme, rather than comparing relative runlengths.
The essence of the scheme is as follows. Consider the pattern shown in the
figure below, which represents a few scanlines from an image. Because each
changing pixel is underneath, or at least very close to, that on the line above,
its position is encoded using a 'Vertical' code. This encoding is used
where the relative positions are within three pixels of each other, and is the
most commonly occurring type of code. In particular, the case where the
positions of the changing pixels are identical, known as V(0) (the others being
VL(3) to VR(3)), is encoded using a single bit. It can be seen immediately that
a vertical line, of any thickness, can be encoded very efficiently.
When the bottom of image is detected (because the position of the changing pixel
is very different to that on the line above), this is encoded as a 'Pass' code,
and represents skipping over the two changing pixels, to black and back to white,
on the line above. The positions of changing pixels which are not in close
proximity to those above are encoded in pairs, using the Group 3 encoding
mechanism.
For 200 DPI (dots per inch) A4 documents, the typical compression ratio of the
CCITT T.6 coding scheme is from about 15:1 to 20:1. For 400 DPI document images
(e.g., eight CCITT test documents), it ranges from 30:1 to 40:1.