Digital Signal Processing Reference
In-Depth Information
image and video, which are needed to conduct on-site video coding and transmit
the stream into center node, decoding, and playing. These devices are relatively
simple, and the operational ability and power itself is very limited. There exist
significant differences between the aspects of power display processing capabilities
and memory support hardware and traditional computing equipments that are far
from being able to meet the high complexity of motion estimation and other
algorithms in traditional video coding, but in the decoding end (such as base stations,
center nodes) have more computing resources and are able to conduct complex
calculations, contrary to application occasions of the traditional video coding. On
the other hand, there exist problems of channel interference, network congestion,
and routing delay in the Internet network that will lead to data error and packet loss,
while the random bit error and unexpected error and other problems in wireless
communication network channel further worsen the channel status, making for a
large number of fields of transmitted video data failure or loss. These problems
are fatal to data compression because the compressed data are generally the stream
consisting of unequally long codes, which will cause error diffusion and other
issues. If there is an error or data packet loss, this will not only affect the service
quality of video business but also cause the entire video communication system to
completely fail and become the bottleneck of restrictions on the development of
real-time network video technology.
From the perspective of video coding, the traditional video coding method, such
as MPEG, H.26X series standard, as a result of using motion estimation, motion
compensation, orthogonal transformation, scalar quantization, and entropy coding
in the coding end, causes higher computational complexity. Motion estimation is
the principle mean to remove the correlation between the video frames, but at the
same time, it is a most complex operation, because every coding block must do
similarity calculations with every block of the reference picture. Comparatively
speaking, in the decoding end, without the search operation of motion estimation, its
complexity is five to ten times easier than the coding end. Therefore, the traditional
video coding method is applied in situations when the coding end has stronger
computational capabilities or the one-time compression and multiple decoding of
non-real time, such as broadcasting, streaming media VOD services, and so on.
On the other hand, the traditional video coding focused more upon improving
the compressive properties, when data transformation error occurred, is mainly
dependent on the correcting capacity of post-channel coding. The international
video coding standard set recently, such as the Fine Granularity Scalability in
MPEG-4 [
9
] and the Progressive Fine Granularity Scalability with higher quality
proposed by Wu Feng [
10
], also tries to adopt a new coding frame to better adapt
to the network transmission. In FGS coding, in order to ensure the reliability of
transmission, the basic layer adopts stronger error protection measures such as the
stronger FEC and ARQ. But the following problems exist in this method: firstly,
the system quality will seriously decline when the network packet loss is serious;
in addition, repeat ARQ will cause excessive delay; and strong FEC will also bring
additional delay because of its complexity, seriously affecting the real-time play of
the video.
