MPEG-4 VIDEO STANDARD VERIFICATION MODEL

1
MPEG-4 VIDEO STANDARD VERIFICATION MODEL

• Thomas Sikora, Senior Member, IEEE

Presenter Chiyoko Kawano
2
Overview

• Introduction
• Functionality
• Toolbox Approach
• Important/New Features
• References

3
Introduction

• MPEG-4 Video Standard Verification Model
  Standardizes algorithms for audio-visual coding
• Requirements
• Interactivity
• High compression
• Universal accessibility and portability of audio
  and video content
• Bit rates between 5 kb/s and 2Mb/s

4
Functionality (1)

• Content-Based Interactivity
• Content-Based Manipulation and Bitstream Editing
• Hybrid Natural and Synthetic Data Coding
• Improved Temporal Random Access

5
Content-Based Manipulation
6
Functionality (2)

• Compression
• Improved Coding Efficiency
• Coding of Multiple Concurrent Data Streams
• Universal Access
• Robustness in Error-Prone Environments
• Content-Based Scalability

7
Multiple Concurrent Data Streams
8
Toolbox Approach (1)

• MPEG-4 standardizes video tools
• Tools algorithms, coding modules, or related
  techniques
• Allows the hybrid coding of Natural (pixel based
  video) and synthetic visual input (i.e. 2-D or
  3-D computer model data set)
• The tools can be flexibly combined at the encoder
  and decoder

9
Toolbox Approach (2)
10
MPEG-4 System Description Language (MSDL)

• Definition of the interfaces between the coding
  tools
• Mechanism to combine tools and to construct
  algorithms and profiles
• Mechanism to download new tools
• MSDL will transmit the structure and rules with
  the bitstream for the decoder

11
Important/New Features (1)

• 420 format
• Quantized into 8 b.
• Image size and shape depends on the application
• Video Object Planes (VOPs)
• Separate coding of Shape, Motion and Texture
• Fixed and variable frame rates
• Enhanced motion estimation and compensation

12
Important/New Features (2)

• Efficient prediction of dc- and ac- coefficients
  of the DCT in intra coded VOPs
• Adaptive macroblock slices for resynchronization
• Backward compatibility with standard H.261/3 or
  MPEG-1/2

13
Macroblocks
14
VOP
15
VOP based decoding
16
The three modes of VOP coding

• P-VOP

17
Shape Coding

• Rectanguler
• No shape information will be coded
• Binary
• 0 Outside of the object
• 1 Inside of the object
• Grayscale (Alpha channel)
• Degree of transparency (0 - 255)

18
Binary Shape Coding
19
Nonoverlapping VOPs
20
Overlapping VOPs
21
Example of Macroblock Grid
22
Padding for Shape (1)
23
Padding for Shape (2)
24
Context Based Arithmetic Codingfor Boundary
Blocks
25
Advanced Motion Prediction
26
Unrestricted Motion Vector
27
Texture Coding
28
Padding for Texture (Intra block)
29
Adaptive DC Prediction

• macroblock

30
Adaptive AC Prediction
31
Spacial Scalability
32
Temporal Scalability (1)
33
Temporal Scalability (2)
34
Temporal Scalability (3)
35
Error Resilience

• Resynchronization
• Insert unique marker
• MPEG-2/H263 After constant number of macroblocks
• MPEG-4 After constant number of bits
• Data partitioning
• Separates the bits for coding of motion
  information and those for the texture information

36
Error Resilience

• Header Extension Code
• Redundant header information, vital for correct
  decoding of video
• Reversible VLCs
• Can decode backwards -gt reduces error portion

37
Resynchronization
38
Reversible VLC
39
References

• Touradj Ebrahimi and Caspar Horne
• "MPEG-4 Natural Video Coding - An Overview"
• http//leonardo.telecomitalialab.com/icjfiles/mpeg
  -4_si/7-natural_video_paper/7-natural_video_paper.
  htm
• Touradj Ebrahimi "MPEG-4 Natural Video Tool "
• http//mpeg.telecomitalialab.com/mtx/mtx97_tebrahi
  mi/
• www.pioneer.co.jp/crdl/tech/mpeg/4-5.html
• Atul Puri and Tsuhan Chen "Multimedia Systems,
  Standards, and Networks"

40
Padding for Shape
41
Padding for Shape
42
DC Prediction