wilamowski-b-m-irwin-j-d-industrial-communication-systems-2011

27-4 Industrial Communication Systems
the image processing application, then 2000 × 2000 pixels will be necessary, captured by a sensor with
a sufficient resolution. This means that a transmission capacity of at least approximately 960.Mbps is
required. In the case of industrial applications where 3D information is needed, this value has to be multiplied
by the number of cameras needed. Due to these high data rates, it is sometimes necessary to use
image compression/decompression, although these operations can introduce latency into the process.
Depending on the effectiveness of the compression and also the limits imposed by the type of application,
this process will produce a compression ratio, normally expressed in X:1, being X the number of bytes
that can be represented using only 1 byte, or in bpp, thus expressing the reduction as well. For example, an
image of 800 × 600 pixels with 24 bpp requires 11,520,000 bits, which is 1.44 Mbyte. If we compress using
a 5:1 ratio, there will be 2,304,000 bits, giving an equivalent compression ratio of 4.8 bpp.
27.2 Multimedia Compression: A Review
The aim of image compression is to reduce the application’s bandwidth requirements (or storage in other
application types) without distorting significantly the original information, although this causes latencies,
as it can be seen in Figure 27.3. Distortion or errors generated by the compression and decompression
process of the image must be limited and have to be measured:
e( x, y) = f ( x, y) − g( x, y)
(27.1)
In Figure 27.3, we can see how the original image must go through a series of steps to achieve a reduction
of the information to be transmitted. This will cause a latency T c (compression time) that will depend on
the complexity of the algorithm used. The compressed information must then be transmitted through
the network, which means it is necessary to packetize the information and wait a period of time that
will depend on the scheduling policy of the source and the priority of the multimedia information with
regard to other information that is using the same medium (e.g., an alarm could have more priority).
Finally, we have to access the medium and carry out the transmission of the information. This process
will introduce a latency T t (transmission time) that will be smaller the more efficient the compression
process has been. There is also an unpack/reception time, but this is almost insignificant, and above all,
a decompression time (T u ) depending on the chosen compressor.
There are several classifications of compressors, depending on the distortion that they introduce in
the source, the complexity of the transmitter and receiver, and the type of redundancy employed. In
the first place, there are compressors with or without losses. Lossless compressors do not distort the
image, that is, f(x, y) = g(x, y). These compressors are not used in excess because the rate of compression
they can achieve is approximately 2:1, which restricts their use in the industrial area. However, they are
used in other types of applications, such as medical applications, where degradation that can distort the
medical diagnosis is not tolerated and where transmission latencies are not critical. In other types of
f (x, y)
Original image
Received image g(x, y)
Compression Transmission Uncompression
Image
preparation
Transformation
domain
Quantization
Encoding
Packetization
Scheduling
Reverse
process
Previous
image
Motion
analysis
Access
medium
Transmission
T c T t T u
FIGURE 27.3
General multimedia application scheme.
© 2011 by Taylor and Francis Group, LLC