| Summary: | 博士 === 國立中央大學 === 資訊工程研究所 === 86 === Transmission of digital video via network has witnessed its
popularity in current Internet applications. With the
standardization of MPEG video coding, visual information
transmission will no doubt play a major role in future Nation
Information Infrastructure (NII). NII supports different
applications, such as Digital-TV (DTV), Video-On-Demand (VOD),
and High-Definition-Television (HDTV) over the inter-connected
networks with different transmission characteristics including
circuit switching and packet switching. Digital video
transmission over the network requires strict timing
synchronization between the server site and client site to
guarantee real-time playback. Therefore, video coding based on
an appropriate bit rate control is a key to the success of
accommodating visual communication in NII environment. In this
dissertation, two extremely efficient bit rate control
algorithms are proposed for different applications: one is
pseudo-constant bit rate video coder which is designed for
source video compression, another is video transcoder which is
designed for scaling the compressed MPEG-2 bit steam. The
proposed technique, Adaptive Scene Analysis (ASA), is applied to
source video bit rate control. ASA intelligently collects
required information to obtain estimated bit count from the
previous, present, and future frames. Combining with a series
of dynamic rate control schemes, the ASA demonstrates
impressive quality transition between successive frames. In
addition, the bit rate irregularities caused by dramatic scene
transition can be smoothed out in a very short period. The
decoded picture quality is consistent among the successive
frames. A comparison with the MPEG-2 Test Model 5 is also
presented in this dissertation. An extremely efficient MPEG-2
video transcoding scheme, Low-Complexity and Low-Delay Video
Transcoding (LCLD) is also proposed in this dissertation to
freely adapt the bit rate of pre-compressed video into
heterogeneous network constraints. The proposed strategy,
LCLD, has low buffer requirement and results in low delay.
Most importantly, the proposed method does not need the
computational intensive motion estimation. Simulation results
demonstrate that the proposed approach results in very
consistent video quality and maintains the buffer level
effectively. Comparing with direct decode/encode approach, the
proposed scheme only suffers slight quality degradation.
Experimental results demonstrate that the proposed approach
can adapt the compressed video bit stream rate to the various
network channel capacities.
|
|---|
