| Summary: | 碩士 === 國立中央大學 === 通訊工程學系 === 106 === Abstract
The current and new video devices and applications require
an efficient compression methods. Streaming of video
to different devices over different types of networks has
substantially increased during the past few years, with
devices ranging from connected televisions to smart-phones
and networks ranging from wired networks to wireless
networks making use of Long Term Evolution Advanced
(LTE Advanced). The rapid development of radio television
and video entertainment industries has increase the demand
for high definition, high frame rate video applications.
As the H.264/AVC mainstream compression standard has
its deficiency, In April 2013, JCT-VT meeting held in Dresden,
Germany, defined the new generation video coding standard
called High Efficiency Video Coding (HEVC). HEVC is
aimed to double the compression efficiency of videos, which
means that the bitrate will be reduced by 50 percent under
the same reconstruction quality. However, when video
data are transmitted through IP networks or wireless mobile
networks, random error and loss of video data always
occur. The fact that HEVC still adopts the same inter
and intra frame prediction mechanism as in H.264, high
compressed video data are prone to channel errors. Sometimes
even few errors may have serious impact on the reconstruction
quality of video data. This paper closely analyses the
characters of HEVC, and further studies error concealment
technology in the post-decoding process. Then it proposes
error concealment algorithms which take the characters
of HEVC into account. An extended hybrid error concealment
algorithm using linear interpolation for error-prone video
transmission channel is been suggested. In the proposed
hybrid algorithm, the selective motion field interpolation (SMFI) is first employed to conceal the erroneous LCU.
The spatial and temporal boundary-matched errors are
then used to check whether the SMFI conceals the erroneous
macroblock properly. If the temporally recovered macroblock
is reconstructed incorrectly, the spatial error concealment
using linear interpolation is employed to conceal the damaged
macroblock instead of SMFI. The proposed hybrid algorithm
employs the linear interpolation on both spatial and temporal
error concealment.
|
|---|
