Game Development Reference
In-Depth Information
be addressed in terms of the human needs, perceptions, feelings and experiences.
The video encoder optimization, such as bit allocation and quality evaluation, can
also be performed with the QoS purpose. In typical video broadcasting system, each
individual has own preference, which may be recorded or learned from the regu-
lar behaviors. From the users experience and preference, an intelligent codec should
have the ability to adaptively perform rate allocation and generate the user experience
preferred video.
In the literature, most of the video coding technologies in literature are performing
optimization on one dimensional, i.e. the best quantization parameter. However, in
addition to the quantization levels, many factors have influence on the visual of expe-
rience, such as spatial and temporal resolution (Zhai et al.
2008
), viewing distance
and even the viewing device. The final quality is determined by these multidimen-
sional factors, and how to jointly consider these factors in encoder optimization is
still an open problem.
12.4.2 Video Compression in Cloud
In the literature, cloud-based image compression techniques has demonstrated its
high efficiency. This can be explained by the reason that intercorrelation in cloud is
exploited to further reduce the redundancy in images, while in conventional image
compression, only spatial correlation is exploited. Yet, different from image compres-
sion, the main compression gain in video coding comes from reducing the interframe
redundancy. Therefore, it is difficult to directly apply cloud image compression tech-
niques to interframe video compression. One opportunity in video compression is
that, distorted frames are used for interprediction, which may enlarge the energy of
the prediction residuals. Therefore, in cloud video coding, similar images and videos
in cloud can be used to reconstruct the video frame within the prediction loop. This
can kill three birds with one stone: generating better quality reference for interpre-
diction, achieving better visual quality reconstruction, and building the correlation
among video frames in cloud.
12.4.3 High Definition Video Coding
The dramatically increasing of high definition (HD) and beyond-HD (e.g., 4
k
×
2
k
or 8
k
4
k
) videos are creating stronger demand of high efficiency video coding
technology. Different from low resolution videos, HD videos have different statistical
properties. In HEVC and AVS2, the coding unit can be adapted to the video content,
and larger unit can be used to compress high resolution video. In Dong et al. (
2009
), it
is found that the HDvideos have higher spatial correlation than low resolution videos,
and thus advanced transform techniques are used for HD video coding. Though
many research works have been done to improve the HD video coding performance,
×
