Video Coding

Although reducing the number of pixels in an image reduces the bit rate of a video
signal quite dramatically, further reductions are necessary if video is to be supported
over the limited bandwidth channels of a cellular network. For most applications,
the number of frames per second can also be reduced from the standard 50
or 60 frames per second used in broadcast systems to a figure between 10 and 30
frames per second. Frame rates as low as 10 frames per second will be acceptable for
some applications, such as videoconferencing or videocalling.

Further reductions in bit rate are achieved by employing coding or compression
techniques. Although there are a number of techniques on the market for processing
video, they have similarities in terms of concept, using a combination of spatial and
temporal compression (Figure 4.36). Spatial compression techniques analyze redundancy
within a frame, produced for example by a large number of adjacent pixels all
having the same or similar levels of brightness (luminance) and color (chrominance).
This redundancy can then be removed by coding. In a similar fashion, temporal
compression looks for redundancy between adjacent frames; this is often the result
of an image background, for example, that does not change significantly between
one frame and the next. Again, this redundancy can be removed by coding.
As the power of electronic processors, particularly digital signal processors
(DSPs), has improved, video codecs have been designed that are able to offer equivalent
quality to their predecessors but at reduced bit rates.

Coding for still images is based on the same spatial compression as used for
video; there is no need to apply temporal compression to a single image. However,
the different still image formats are better suited to one type of image or another.
For example, JPEG works well with black and white or color natural images (such
as photographs), whereas GIF works better for black and white images that contain
lines and blocks (such as cartoons).

There is also a distinction between coding that is lossy and coding that is lossless.
The video coding techniques described here and JPEG for still images are all
classified as lossy in that they remove information through coding that cannot be
regenerated later. On the other hand, GIF is a lossless coding technique and subsequently
does not remove information through its coding.

Video Coding Standards
A number of video coding standards exist in commercial applications and two
main groups have worked on these standards: (1) the International Standards Organization
(ISO) and (2) the International Telecommunication Union – Telecommunications
branch (ITU-T). The ISO is responsible for the Moving Pictures Expert
Group (MPEG) that has produced a series of video coding systems (Figure 4.37).
The first MPEG standard, MPEG-1, was released in 1992 and was aimed at
providing acceptable, but sub-broadcast, quality video that could be used for CDs
and games. The video coding produced an output at 150 kbps. The audio coding
portion of the MPEG-1 standard included three coding options offering progressively
greater compression rates for the audio component. The third of these
options, MPEG-1 Layer 3 or simply MP3, has become a dominant standard for the
distribution of music and audio over the Internet.

In 1994, MPEG-2 was released, and offered improvements on the MPEG-1
standard. MPEG-2 can work at a variety of bit rates, but at 1 to 3 Mbps outperforms
the quality of MPEG-1. MPEG-2 is used for digital versatile disks (DVDs)
and digital video broadcast (DVB), and includes a range of audio coding options
such as advanced audio coding (AAC).

The most recent addition to the MPEG family, MPEG-4 was originally designed
for low bit services, with channels operating sub-64 kbps but can also be employed at
high bit rates into the Mbps range. The design intention was to provide video coding
for some of the “new” applications that were appearing, such as streaming services