Mapping GOPS in an Improved DVC to H.264 Video Transcoder - Advanced Concepts for Intelligent Vision Systems - page 24

Fig. 2. Proposed WZ to H.264 Video Transcoder
transcoder is allocated in the network where more resources are allowed. The
architecture of the proposed transcoder is depicted in Figure 2 and works as follows:
the WZ encoder is based on VISNET-II architecture [7] which is an evolution of
DISCOVER architecture [8]. In this work we have used Transform Domain (TD). At
the encoder, K frames are coded using H.264 Intra and WZ frames are coded
following the basis of WZ paradigm [2]. In our experiments we are working with WZ
GOP sizes 2, 4 and 8, but the transcoder could accept every GOP length as will be
explained in section 3.1. In the second half of the transcoder, there is a H.264 encoder
which converts the WZ output into a H.264 bitstream using a GOP I11P because it is
the most suitable pattern for mobile-to-mobile video communications. To develop this
conversion, every K frame which matches with an I-frame is passed without any
transcoding process. On the other hand, every K or WZ frame which matches with a P
frame is encoding using the method proposed to reduce the ME time which is
explained in section 3.2.
3.1 Mapping GOP Patterns
Side information generation is a crucial task for any WZ codec due to the fact that
WZ frames are decoded and reconstructed departing from the SI. There are many