Mapping GOPS in an Improved DVC to H.264 Video Transcoder - Advanced Concepts for Intelligent Vision Systems - page 22

appeared in the multimedia community as a solution for this mobile to mobile video
communication, as it is shown in figure 1. In this framework the majority of the
computation is moved to the network where the transcoder is allocated and the
simpler algorithms (DVC encoding and H.26X decoding) are implemented in the end
user devices. At the moment transcoding architecture offers the most suitable solution
for mobile-to-mobile video communications due to the low complexity in both
extremes. In the literature the Group of Pictures (GOP) pattern in the DVC
architectures is formed by two kinds of frames: Wyner-Ziv (WZ) and Key (K).
Normally with GOP sizes ranging from 2, 4 and 8 (WZ frames between two K
frames) although other GOP sizes are also allowed. On the other hand, in H.264 the
most suitable GOP pattern for mobile communications is labeled as I11P [5] which is
formed by one I-frame followed by 11 P-frames and with GOP size of twelve. These
GOP sizes / patterns mismatches between DVC and H.264 entail a problem that must
be solved in the proposed scenario.
Fig. 1. System using a DVC / H.26X transcoder
At this point, this paper is a straight forward step in the framework of DVC to
H.264 video transcoders and offers a GOP mapping solution between kinds of GOP
sizes / patterns as well as some refinement in the motion estimation algorithm
developed at H.264 encoding algorithm as part of the whole video transcoder to make
a faster process. Accelerating the transcoding process is a very important task in order
to try to reach a real time communication.
This paper is organized as follows: Section 2 identifies the state-of-the-art in DVC
based transcoders. Section 3 shows the proposed mapping algorithm for the DVC to
H.264 video transcoder which is evaluated in Section 4 with some simulation results.
Finally, in Section 5, conclusions are presented.