Transmission - 3DTV: Processing and Transmission of 3D Video Signals - page 99

be achieved explicitly by embedding duplicates of the primary slices as
redundant slices, or implicitly by means of either Reversible Variable Length
Coding (RVLC) or Multiple Descriptive Coding (MDC).

• Redundant coding - In this tool, the replicas of the primary coded pictures

(PCP) will be encoded and transmitted along with the primary data
stream. At the decoder the redundant pictures are decoded if and only if
the primary pictures are lost, otherwise the redundant stream is discarded.
Therefore, the redundant coding is extremely disadvantageous in error-
free environments. Hence, measures have been taken to minimize the
amount of channel bandwidth allocated for the redundant frames.

• Reversible Variable Length Coding - RVLC provides error resilience by means

of both forward and backward decoding techniques. In other words, if an
error (E) takes place between two adjacent synchronization markers (SM)
in a bit stream, in general the whole data between the two markers will
be discarded, regardless of the accuracy of the rest of the data. However,
this can be overcome by the use of RVLC, which has the capacity to
carry out decoding in both forward and backward directions as shown
in Figure 4.2. The code words of the Variable Length Coder (VLC) must
be symmetric, so that it can be decoded either in the forward direction or
in the backward direction. This in turn limits the flexibility of VLC and
decreases the efficiency of the codec.

• Multiple Descriptive Coding (MDC) - In this technique the bit stream is

divided into mainly two streams which can be decoded independently to
give a reasonable quality output video. The two streams are transmitted
either in two different channels or multiplexed into one channel. At the
decoder a reasonable quality image sequence can be regenerated if one
of the streams is received without any errors. Moreover, if both streams
arrive at the receiving end without any errors, then the decoder is able to
generate an output image sequence with superior quality. Furthermore,
the redundancy in the transmission data can be minimized by analysing
the correlation presence between the two streams, at the encoder.

4.2.2.4 Concealment Driven Tools

As the name implies, in concealment driven tools the errors are concealed by
means of the additional information received about the general behaviour
of the lost data. These tools can be either encoder driven or decoder driven.
Interpolation is a common method which has been used for the lost data
prediction.
Examples of error concealment methods used in AVC are:

1. Frame copy - In this algorithm the error is concealed by copying the first
picture in the reference list 0 of the corresponding sample. It can be
performed on both base layer and enhancement layer levels.