Compression - 3DTV: Processing and Transmission of 3D Video Signals - page 88

coding standard (MVC) that had already attracted many researchers and
multimedia technology companies worldwide. The MVC standard proved to
be a success and superior to the native AVC standard as regards stereoscopic
or multi-view compression. From the coding efficiency point of view, as the
number of possible representations of a certain scene increases, i.e. different
viewing angles, the amount of bit-rate budget to represent the multi-view
video increases and hence, more sophisticated methods should be employed
to compress the content. Future 3DTV systems will require multi-view coding
schemes, which are able to compress the 3D multi-view video content to be
transmitted over band-limited networks under heavy load, while providing
interoperability and scalability at the same time.
On the other hand, the current MVC approach is conceptually unable to
keep the 3D video transmission load at manageable rates, while the number
of camera views to shoot a particular scene keeps increasing for higher
3D resolution (targeting multi-view displays). Hence, it is not possible to
commercially realize 3DTV systems based on multi-view video services
that consist of extremely many cameras and use the current compression
technology. Table 3.1 outlines the estimate source coding bit-rate for various
3D video formats (at HD resolution and 25 fps, progressive coding) with
three state-of-the-art video coding standards. Note that HEVC is planned be
standardized in early 2013 following the development work. It can be seen
that as long as the number of camera views involved in the 3D video format
increases, the bit-rate of all encoders tends to increase linearly.
Based on the rate constraint, a newer 3D video framework has been intro-
duced by the standardization body MPEG that puts a limit on the input
number of cameras [45]. The 3D video framework suggests ideally coding
two viewpoints or at most three viewpoints and sufficiently producing as
many viewpoints as needed at the client (decoder) side. This suggestion is
also in line with the limitations applicable in the widely deployed mod-
ern production platforms, which can accommodate a restricted number of
cameras in a compact, calibrated and easily manageable manner. The main
purpose is to set an upper limit on the required bit-rate to transmit the
corresponding 3D video format. In other words, for a given resolution, as

Table 3.1 Indicative bit-rates necessary to transmit various MVV formats
using state-of-the-art video coding standards

Video Resolution@ 2D Stereo- Four Eight

Codec Frame-rate (One view) scopic Views Views

(Pixels@ (Mbit/s) (Mbit/s) (Mbit/s) Mbit/s

Frames/second)

AVC 1920 × 1080 p @25 10-20 20-40 40-80 80-160

MVC 1920 × 1080 p @25 10-20 14-28 28-56 56-112

HEVC 1920 × 1080 p @25 5-10 10-20 20-40 40-80