Compression - 3DTV: Processing and Transmission of 3D Video Signals - page 68

investigated by the Motion Picture Experts Group (MPEG) in the early
1990s. The first generation of MPEG, MPEG-1 coding standard, uses H.261
as a starting point. MPEG-1 optimally operates in the range of 1.2 - 1.5 Mbps
for non-interlaced video. Later, a new generation of standards emerged for
coding interlaced video at higher bit-rates (in the range of 4 - 9 Mbps) called
MPEG-2. MPEG-2 became so popular that it is used in many digital video
applications such as terrestrial video broadcasting (DVB-T), satellite TV,
cable TV and Digital Versatile Disc (DVD).
Later, MPEG-2 was adopted by ITU-T under the generic name of H.262.
After progressive developments on MPEG-1 and MPEG-2, MPEG started
working on a video-object-based coding standard and came up with MPEG-4.
ITU-T, in parallel, carried out some work on a new video coding standard
H.263 targeting coding at very low bit-rates. The evolutions of this standard
were called H.263 , where the compression efficiency of this

video coding standard was improved over the years. In 1997, two groups,
ITU-T and ISO/IEC MPEG came together and formed a Joint Video Team
(JVT) in order to create a single video coding standard: H.26L. Later, this codec
was published jointly as Part 10 of MPEG-4 and ITU-T Recommendation
H.264 [1] and called Advanced Video Coding (AVC). AVC is among the latest
entries in the series of international coding standards [1]. AVC has achieved
a significant improvement in rate-distortion efficiency relative to existing
standards. It is noteworthy that AVC has almost double the compression
performance of MPEG-2. In the following subsection, a brief overview of
AVC is given.

+ and H.263 ++

3.2.1 OverviewofMPEG-4Part10/H.264AVCStandard

Described by its 'network-friendly' characteristic and high rate-distortion
performance over heterogeneous networks, AVC is used in quite a wide
range of video coding applications including broadcast over various media,
multimedia data storage; conversational, multimedia streaming and mul-
timedia messaging services on Ethernet, LAN, DSL, wireless and mobile
networks, modems, etc. Figure 3.2 shows the high-level encoder block
diagram of the AVC coding standard.
The frames of video are split into 16

16 pixel blocks called macroblocks.
The macroblocks are encoded in a raster scan order from top left to bottom
right of the frame. Every macroblock inside a single frame consist of three
components: Luminance (Y), and two chrominance components Cr and Cb
(representing colour). The chrominance samples are usually sub-sampled by
a factor of two in both horizontal and vertical directions, since the human
visual system is less sensitive to chrominance channels. Hence, a macroblock
consists of 16

×

8 chrominance samples. The
macroblocks are grouped as slices and there are five types of slices defined
in AVC, which are Intra (I), Predictive (P), Bi-predictive (B), Switch-I (SI) and
Switch-P (SP) slices.

× 16 luminance and two 8 ×