CHAPTER 9 H.264MPEG4 Part 10

1
CHAPTER 9H.264/MPEG-4 Part 10
2
H.264 The Emerging Video Coding Standard
3
Basic Macroblock Coding Structure
4
Profiles and Levels

• H. 264 defines a set of three Profiles
• Baseline Profile
• Main Profile
• Extended Profile
• Baseline Profile
• Support intra and inter-coding (using I-slices
  and P-slices) and entropy coding with
  context-adaptive variable-length codes (CAVLC).
• Potential applications of Baseline Profile
  include videotelephony, videoconferencing and
  wireless communications.

5
Profiles and Levels (Cont.)

• Main Profile
• Support for interlaced video, inter-coding using
  B-slices, inter coding using weighted prediction
  and entropy coding using context-based arithmetic
  coding (CABAC).
• Potential applications include television
  broadcasting and video storage
• Extended Profile
• Does not support interlaced video or CABAC but
  add modes to enable efficient switching between
  coded bitstreams (SP- and SI-slices) and improved
  error resilience (data partitioning)
• Streaming media applications

6
Profiles and Levels (Cont.)
B Slices
Interlace
SP and SI Slices
Weighted Prediction
CABAC
Data Partitioning
Main Profile
I Slices
P Slices
Extended Profile
CAVLC
Slice Groups And ASO
Redundant Slices
Baseline Profile
7
Video Format

• H.264 supports coding and decoding of 420
  progressive or interlaced video.
• An H. 264 encoder may use one or two of a number
  of previously encoded pictures as a reference for
  motion-compensated prediction of each inter coded
  macroblock or macroblock partition.
• Motion Compensation
• Seven kinds of block sizes 16 ? 16, 16 ? 8, 8 ?
  16, 8 ? 8, 8 ? 4, 4 ? 8, 4 ? 4
• Multiple reference pictures

8
Coded Data Format

• H. 264 makes a distinction between a Video Coding
  Layer (VCL) and a Network Abstraction Layer
  (NAL).
• The output of the encoding process is VCL data
  which are mapped to NAL units prior to
  transmission or storage.
• A coded video sequence is represented by a
  sequence of NAL units that can be transmitted
  over a packet-based network or a bitstream
  transmission link or stored in a file.

9
Multiple Reference Frames
10
Variable Block Size
11
Example of Variable Block Size
12
Motion Vectors

• Each Partition or sub-macroblock partition in an
  inter-coded macroblock is predicted from an area
  of the same size in a reference picture.
• The offset between the two areas (the motion
  vector) has quarter-sample resolution for the
  luma component and one-eighth-sample resolution
  for the chroma components.
• If one or both vector components are fractional
  values, the prediction samples are generated by
  interpolation between adjacent samples in the
  reference frame.

13
Interpolation of Luma Half-Pixel Positions
14
Interpolation of Luma Quarter-Pel Positions

• Once all the half-pel samples are available, the
  samples at quarter-pex positions are produced by
  linear interpolation.
• a round((G b)/2)

15
Interpolation of Chroma Eighth-Sample Positions

• Quarter-pex resolution motion vectors in the luma
  component require eighth-pel resolution vectors
  in the chroma components (assuming 420
  sampling).

a round(8 - dx)8 - dy)A dx(8 - dy)B
(8 - dx)dyC dxDyD/64) In this
example Dx is 2, and dy is 3, so that a
round(30A 10B 18C 6D)/64
B
A
dy
dx
8 - dx
A
8 - dy
C
D
16
Motion Vector Prediction

• Encoding a motion vector for each partition can
  cost a significant number of bits, especially if
  the small partition sizes are chosen.
• Motions vectors for neighbouring partitions are
  often highly correlated and so each motion vector
  is predicted from vectors of nearby, previously
  coded partitions.
• A predicted vector, MVp, is formed based on
  previously calculated motion vectors and MVD, the
  difference between the current vector and the
  predicted vector, is encoded and transmitted.

17
Current and Neighbouring Partitions
C 16 ? 8
B 4 ? 8
A 8 ? 4
E 16 ? 16
18
Intra Prediction

• In intra mode a prediction block P is formed
  based on previously encoded and reconstructed
  blocks and is subtracted from the current block
  prior to encoding.
• For the luma samples, P is formed for each 4 ? 4
  block or a 16 ? 16 block.
• There are a total of nine optional prediction
  modes for each 4 ? 4 luma block, four modes for a
  16 ? 16 block.
• One macroblock mode for chroma
• Similar to intra 16 ? 16.

19
4 ? 4 Luma Prediction Modes
20
4 ? 4 Luma Prediction Modes (Cont.)

• Mode 2 DC Prediction
• Mode 0-8 (except 2) direction prediction

21
DC Prediction
22
Mode 0 and Mode 1
23
Mode 3 and Mode 4
24
Mode 5 and Mode 6
25
Mode 7 and Mode 8
26
16 ? 16 Luma Prediction Modes
27
Intra 16 ? 16 Plane Prediction
28
Bitrate Saving (Real Time)
29
Bitrate Saving (Storage)
30
Objective Evaluation
31
Objective Evaluation (Cont.)
32
Subjective Evaluation
33
Computation Complexity of H. 264
34
Analysis, Fast Algorithm, and VLSI Architecture
Design for H.264/AVC Intra Frame Coder

• Yu-Wen Huang, Bing-Yu Hsieh, Tung-Chien Chen, and
  Liang-Gee Chen, Fellow, IEEE
• IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR
  VIDEO TECHNOLOGY, VOL. 15, NO. 3, MARCH 2005

35
Core Techniques Adopted in Different Image Coding
Standards
36
Comparisons Between Different Image Coding
Standards
37
Intraprediction Modes

• In H.264/AVC intra coding, two intra-macroblock
  modes are supported. One is intra 4 4 prediction
  mode, denoted as I4MB, and the other is intra 16
  ? 16 prediction mode, denoted as I16MB.

I4MB left illustration of nine 44-luma
prediction modes right examples of real images.
38
16 ?16 Luma Prediction Modes
39
Transform and Quantization

• H.264/AVC still adopts transform coding for the
  prediction error signals. The block size is
  chosen as 4 ? 4 instead of 8 ? 8.
• The 4 ? 4 DCT is an integer approximation of
  original floating point DCT transform. An
  additional 2 ? 2 transform is applied to the four
  dc-coefficients of each chroma component.
• If a-luma macroblock is coded as I16MB, two
  dimensional (2-D) 4 ? 4 Hadamard transform will
  be further applied on the 16 dc coefficients
  of-luma blocks.

40
Transform and Quantization (Cont.)

• H.264/AVC uses scalar quantization without
  dead-zone. The quantization parameter QP ranges
  from 0 to 51.
• An increase in QP approximately reduces 12.5 of
  the bit-rate.
• The quantized transform coefficients of a block
  are scanned in a
• zigzag fashion. The decoding process can be
  realized with only
• additions and shifting operations in 16-b
  arithmetic. For more
• details, please refer to 13.
• In H.264/AVC, two entropy coding schemes are
  supported.
• One is VLC-based coding, and the other is
  arithmetic coding