Page 109 - Video Coding for Mobile Communications Efficiency, Complexity, and Resilience

86 Chapter 3. Video Coding: Standards

VOPs are input to the scalability preprocessor. If spatial scalability is to
be performed, then this preprocessor downsamples the input VOPs to generate
the base-layer VOPs forming the input to the base-layer encoder. The mid-
processor takes the reconstructed base-layer VOPs and upsamples them. The
di1erence between the original VOPs and the output of the midprocessor forms
the enhancement-layer VOPs. Those are encoded using the enhancement-layer
encoder. The multiplexer is then used to multiplex the base- and enhancement-
layer bitstreams into a single bitstream. At the decoder, the demultiplexer
is used to separate the incoming bitstream into base- and enhancement-layer
bitstreams. The scalability postprocessor performs any necessary operations,
such as upsampling the decoded base layer for display.
If, however, temporal scalability is to be performed, then the scalability
preprocessor separates the stream of input VOPs into two substreams. One
substream forms the input to the base-layer encoder, while the other forms
the input to the enhancement-layer encoder. In this case, the midprocessor
does not perform any spatial resolution conversion and simply allows the
reconstructed base-layer VOPs to pass through to be used for the temporal
prediction of enhancement-layer VOPs. In this case also, the postprocessor
simply outputs the reconstructed base-layer VOPs without any conversion.
For spatial scalability, only rectangular VOPs are supported by MPEG-4. In
the case of temporal scalability, however, both rectangular and arbitrary-shaped
VOPs are supported. MPEG-4 provides two types of temporal scalability:

• Type I: The enhancement layer increases the temporal resolution of only
a partial region of the base layer.

• Type II: The enhancement layer increases the temporal resolution of
the entire region of the base layer.

3.5.8 Error Resilience
One of the main aims of MPEG-4 is to provide universal access through a wide
range of environments, including error-prone environments. One of the impor-
tant requirements of video communication over error-prone environments, like
mobile networks, is robustness against errors. MPEG-4 provides three main
tools for error resilience: resynchronization, data partitioning, and reversible
VLCs.

3.5.8.1 Resynchronization
As is discussed in Chapter 9, one of the disadvantages of VLC coding is
that errors in the bitstream can cause a loss of synchronization between the
encoder and the decoder. One way to reduce this e1ect is to insert unique