SMPTE ST 2042-1-2012 VC-2 Video Compression.pdf

1、 Copyright 2012 by THE SOCIETY OF MOTION PICTURE AND TELEVISION ENGINEERS 3 Barker Avenue, White Plains, NY 10601 (914) 761-1100 Approved August 2, 2012 SMPTE ST 2042-1:2012 Revision of SMPTE ST 2042-1:2009 SMPTE STANDARD VC-2 Video Compression Page 1 of 137 pages Table of Contents Page Foreword . 8

2、 Intellectual Property . 8 Introduction 8 1 Scope . 10 2 Conformance Notation . 10 3 Normative References 10 4 Definition of Acronyms and Terms . 11 4.1 Acronyms . 11 4.2 Terms 11 5 VC-2 Conventions 14 5.1 Naming Conventions: . 14 5.2 State Representation 14 5.3 Number Formats . 14 5.4 Data Types 14

3、 5.4.1 Elementary Data Types. 14 5.4.2 Compound Data Types . 15 5.5 Functions and Operators 15 5.5.1 Assignment . 15 5.5.2 Boolean Functions and Operators 16 5.5.3 Integer Functions and Operators 16 5.5.4 Array and Map Functions and Operators 18 5.5.5 Precedence and Associativity of Operators 18 5.6

4、 Pseudocode 19 5.6.1 Processes and Functions 19 5.6.2 Variables . 20 5.6.3 Control Flow 21 6 Overall VC-2 Specification . 22 SMPTE ST 2042-1:2012 Page 2 of 137 pages 7 Video Formats 24 7.1 Color Model 24 7.2 Interlace 24 7.3 Component Sampling . 25 7.4 Bit Resolution and Signal Ranges 25 7.5 Video F

5、rame Size and Rate . 25 8 Encoding Overview (Informative) . 25 8.1 Picture Input Processing . 26 8.2 Wavelet Transform . 26 8.3 Division Into Subbands . 26 8.4 DC Subband Prediction 28 8.5 Coefficient Scanning . 28 8.5.1 Core Syntax Coefficient Scanning 28 8.5.2 Low Delay Coefficient Ordering and Sc

6、anning . 29 8.6 Quantization and Quantizer Estimation 30 8.6.1 Quantization of the DC band . 31 8.7 Data Encoding 31 8.7.1 Arithmetic Coding 31 8.8 Stream Syntax 32 9 Decoding Overview. 32 9.1 Decoding Functions 32 9.1.1 Functional Description . 33 9.1.2 Data Decoding . 33 9.1.2.1 VC-2 Data Codings

7、. 34 9.1.3 VC-2 Syntax Decoding 34 9.1.4 Subband Decoding 34 9.1.4.1 DC Band Prediction 35 9.1.5 Inverse Quantization . 35 9.1.5.1 Quantizer Factor and Offset (Informative) 35 9.1.6 Coefficient Coding Order . 36 9.1.7 Inverse Discrete Wavelet Transform . 36 9.1.7.1 Wavelet Filter Support 37 9.1.8 Cl

8、ipping . 37 10 VC-2 Stream . 38 10.1 Pseudocode 38 10.2 VC-2 Stream Syntax . 38 10.3 VC-2 Sequence Syntax 38 10.3.1 Parse Info Headers . 39 10.3.2 Data Units 40 10.3.2.1 Auxiliary Data 40 10.3.2.2 Padding . 40 10.4 Parse Info Header Syntax . 41 10.4.1 Parse Codes 42 SMPTE ST 2042-1:2012 Page 3 of 13

9、7 pages 10.4.1.1 Parse Code Values (Informative) . 43 10.5 VC-2 Sequence Decoding (Informative) . 43 10.5.1 Non Sequential Picture Decoding (Informative) 44 11 Sequence Header 45 11.1 Parse Parameters . 47 11.1.1 Version Number 47 11.1.2 Profiles and Levels 48 11.2 Base Video Format . 48 11.3 Source

10、 Parameters . 49 11.3.1 Setting Source Defaults 50 11.3.2 Frame Size 50 11.3.3 Color difference Sampling Format 51 11.3.4 Scan Format 52 11.3.5 Frame Rate . 52 11.3.6 Pixel Aspect Ratio . 54 11.3.7 Clean Area 56 11.3.8 Signal Range . 56 11.3.9 Color Specification 58 11.3.9.1 Color Primaries . 59 11.

11、3.9.2 Color Matrix 59 11.3.9.3 Transfer Function . 60 11.4 Picture Coding Mode 61 11.5 Initializing Coding Parameters 61 11.5.1 Picture Dimensions . 61 11.5.2 Video Depth 62 12 Picture Syntax 62 12.1 Picture Header 63 12.2 Wavelet Transform . 63 12.3 Transform Parameters 64 12.3.1 Wavelet Filter 64

12、12.3.2 Transform Depth . 65 12.3.3 Codeblock Parameters (Core Syntax Only) 65 12.3.4 Slice Coding Parameters (Low Delay Syntax Only) . 67 12.3.4.1 Slice Parameters 67 12.3.4.2 Quantization Matrices . 67 13 Transform Data Syntax 68 13.1 Subband Data Structure . 69 13.1.1 Wavelet Data Initialization .

13、 69 13.1.2 Wavelet Subband Dimensions 70 13.2 Inverse Quantization . 71 13.2.1 Quantization Factors and Offsets . 71 13.3 DC Subband Prediction 72 13.4 Core Syntax Wavelet Coefficient Unpacking 73 SMPTE ST 2042-1:2012 Page 4 of 137 pages 13.4.1 Core Syntax Transform Data 73 13.4.2 Subbands 74 13.4.2

14、.1 Zero Subband . 74 13.4.2.2 Non-Skipped Subbands 75 13.4.3 Subband Codeblocks 75 13.4.3.1 Codeblock Dimensions . 75 13.4.3.2 Codeblock Unpacking Loop 76 13.4.3.3 Skipped Codeblock Flag . 76 13.4.3.4 Codeblock Quantizer Offset . 77 13.4.4 Subband Coefficients 77 13.4.4.1 Zero Parent . 78 13.4.4.2 Z

15、ero Neighborhood . 79 13.4.4.3 Sign Prediction 79 13.4.4.4 Coefficient Context Selection . 80 13.5 VC-2 Low Delay Wavelet Coefficient Unpacking 82 13.5.1 Overall Process . 82 13.5.2 Slice Unpacking for Low Delay Pictures . 83 13.5.2.1 Determining the Number of Bytes in a Low Delay Picture Slice 84 1

16、3.5.3 Slice Unpacking for High Quality Pictures. 85 13.5.4 Setting Slice Quantizers 86 13.5.5 Slice Subbands . 86 13.5.5.1 Slice Subband Area . 86 13.5.5.2 Single Component Slice Subband Data . 87 13.5.5.3 Color difference Slice Subband Data . 87 14 Picture Decoding 87 14.1 Overall Picture Decoding

17、Process 87 14.2 Picture IDWT . 88 14.3 Component IDWT . 88 14.3.1 Vertical and Horizontal Synthesis . 89 14.3.2 One-Dimensional Synthesis 90 14.3.2.1 Mathematical Formulation of Lifting Processes (Informative) . 92 14.3.3 Lifting Filter Parameters 93 14.3.4 Removal of IDWT Pad Values 96 14.4 Picture

18、 Output Ranges 96 Annex A VC-2 Data Coding Definitions (Normative) 98 A.1 Bit Packing and Data Input . 98 A.1.1 Reading a Byte . 98 A.1.2 Reading a Bit 98 A.1.3 Byte Alignment 98 A.2 Fixed Length Data 99 A.2.1 Boolean . 99 A.2.2 n-bit Unsigned Integer Literal 99 A.2.3 n-byte Unsigned Integer Literal

19、 . 99 SMPTE ST 2042-1:2012 Page 5 of 137 pages A.3 Variable-Length Codes . 100 A.3.1 Data Input for Bounded Block Operation . 100 A.3.2 Unsigned Interleaved Exp-Golomb Codes 101 A.3.3 Signed Interleaved Exp-Golomb Codes 103 A.4 Parsing of Arithmetic Coded Data 104 A.4.1 Context Probabilities . 104 A

20、.4.2 Arithmetic Decoding of Boolean Values . 105 A.4.3 Arithmetic Decoding of Integer Values . 105 A.4.3.1 Binary Coding and Contexts 105 A.4.3.2 Unsigned Integer Decoding . 106 A.4.3.3 Signed Integer Decoding . 106 Annex B Arithmetic Coding (Normative) 107 B.1 Arithmetic Coding Principles (Informat

21、ive) 107 B.1.1 Interval Division and Scaling 107 B.1.2 Finite Precision Arithmetic 108 B.1.3 Symbol Probability Estimation 108 B.2 Arithmetic Decoding Engine (Normative) 110 B.2.1 State and Contexts . 110 B.2.2 Initialization . 110 B.2.3 Data Input . 111 B.2.4 Decoding Boolean Values 111 B.2.5 Renor

22、malization 112 B.2.6 Updating Context Probabilities . 112 B.3 Arithmetic Encoding (Informative) . 114 B.3.1 Encoder Variables 114 B.3.2 Initialization . 114 B.3.3 Encoding Binary Values . 114 B.3.3.1 Scaling the Interval 114 B.3.3.2 Updating Contexts . 115 B.3.3.3 Renormalization and Output 115 B.3.

23、4 Flushing the Encoder 115 Annex C Predefined Video Formats (Normative) . 117 Annex D Profiles and Levels (Normative) . 121 D.1 Profiles 121 D.1.1 Low Delay Profile 121 D.1.2 Simple Profile . 121 D.1.3 Main Profile . 122 D.1.4 High Quality Profile . 122 D.2 Levels . 123 Annex E Low Delay Quantizatio

24、n Matrices (Normative) . 124 E.1 Default Quantization Matrices 124 E.2 Quantization Matrix Design and Quantizer Selection (Informative) 125 E.2.1 Noise Power Normalization 125 E.2.2 Custom Quantization Matrices . 127 SMPTE ST 2042-1:2012 Page 6 of 137 pages Annex F Video Systems Model (Informative)

25、. 129 F.1 Color Models . 129 F.1.1 YCBCRCoding . 129 F.1.2 YCOCGCoding . 129 F.1.3 Signal Range . 129 F.1.4 Color Primaries . 130 F.1.5 Color Matrix . 130 F.2 Transfer Characteristics . 130 F.2.1 TV Transfer Characteristic 130 F.2.2 Extended Color Gamut 131 F.2.3 Linear 131 F.3 Frame Rate . 131 F.4

26、Aspect Ratios And Clean Area . 131 F.4.1 Pixel Aspect Ratio . 131 F.4.1.1 Using Non-Square Pixel Aspect Ratios 131 F.4.2 Clean Area. 132 Annex G Wavelet Decimation and Reconstruction Processes (Informative). 133 G.1 Overview of Wavelet Processing 133 G.2 The Lifting Process . 135 Annex H Bibliograph

27、y (Informative) 137 SMPTE ST 2042-1:2012 Page 7 of 137 pages Figures Page Figure 8.1 Sample encoder functional block diagram 25 Figure 8.2 A single DWT transform stage 27 Figure 8.3 Identification of the subbands of a 4-level wavelet transform . 28 Figure 8.4 Division of subbands into codeblocks, il

28、lustrating the constant codeblock structure at each level 29 Figure 8.5a Formation of a single 4x4 slice from 2-level transform coefficients 30 Figure 8.5b An 8x4 array of slices for a 2-level wavelet transform 30 Figure 8.6 Arithmetic encoding process . 32 Figure 9.1 Functional VC-2 decoder block d

29、iagram . 33 Figure 9.2 Prediction aperture for DC bands . 35 Figure 9.3 Wavelet Decoding Steps . 37 Figure 10.1 VC-2 stream 38 Figure 10.2 Overview of VC-2 sequence structure 39 Figure 10.3 Parse Info header syntax 41 Figure 11.1 Sequence Header . 46 Figure 11.2 Parse Parameters . 47 Figure 11.3 Sou

30、rce Parameters . 50 Figure 11.4 Frame Size 51 Figure 11.5 Sampling Format . 51 Figure 11.6 Scan format . 52 Figure 11.7 Frame rate . 53 Figure 11.8 Pixel Aspect Ratio . 55 Figure 11.9 Clean area . 56 Figure 11.10 Signal range 57 Figure 11.11 Color specification syntax . 58 Figure 12.1 VC-2 picture 6

31、3 Figure 12.2 Wavelet transform data . 64 Figure 12.3 Transform parameters. 64 Figure 12.4 Spatial partitioning (core syntax only) . 66 Figure 12.5 Slice parameters . 67 Figure 13.1 4-Level subband array 69 Figure E.1 Subband weights for a 1-level decomposition 126 Figure E.2 Subband weights for a 2

32、-level decomposition 127 Figure G.1 Single wavelet processing stage comprising decimation and reconstruction filters 133 Figure G.2 Illustration of the alias frequency generation and cancellation in a wavelet filter bank . 134 Figure G.3 Two-step wavelet processing filter bank 134 Figure G.4 Decompo

33、sition of a single image into 7 wavelet frequency bands 135 Figure G.5 Decomposition of the EBU “Boats” picture into 7 wavelet frequency bands 135 Figure G.6 Polyphase representation of wavelet filter banks . 136 SMPTE ST 2042-1:2012 Page 8 of 137 pages Foreword SMPTE (the Society of Motion Picture

34、and Television Engineers) is an internationally recognized standards- developing organization. Headquartered and incorporated in the United States of America, SMPTE has members in over 80 countries on six continents. SMPTEs Engineering Documents, including Standards, Recommended Practices and Engine

35、ering Guidelines, are prepared by SMPTEs Technology Committees. Participation in these Committees is open to all with a bona fide interest in their work. SMPTE cooperates closely with other standards-developing organizations, including ISO, IEC and ITU. SMPTE Engineering Documents are drafted in acc

36、ordance with the rules given in Part XIII of its Operations Manual. SMPTE ST 2042-1 was prepared by Technology Committee 10E. Intellectual Property At the time of publication no notice had been received by SMPTE claiming patent rights essential to the implementation of this Standard. However, attent

37、ion is drawn to the possibility that some of the elements of this document may be the subject of patent rights. SMPTE shall not be held responsible for identifying any or all such patent rights. Introduction This section is entirely informative and does not form an integral part of this document. Th

38、e VC-2 standard specifies the compressed stream syntax and reference decoder operations for a video compression system. VC-2 is an intra frame video compression system aimed at professional applications that provides efficient coding at many resolutions including various flavors of CIF, SDTV and HDT

39、V. VC-2 utilises wavelet transforms that decompose the video signal into frequency bands. The codec is designed to be simple and flexible, yet be able to operate across a wide range of resolutions and application domains. The system provides the following capabilities: Multi-resolution transforms. D

40、ata is encoded using the wavelet transform, and packed into the bitstream subband by subband. High compression ratios result in a gradual loss of resolution. Lower resolution output pictures can be obtained by extracting only the lower resolution data. Frame and field coding. Both frames and fields

41、can be individually coded. CBR and VBR operation. VC-2 permits both constant bit rate and variable bit rate operation. For low delay pictures, the bit rate will be constant for each area (VC-2 slice) in a picture to ensure constant latency. Variable bit depths. 8, 10, 12 and 16 bit formats and beyon

42、d are supported. Multiple color difference sampling formats. 444, 422 and 420 video are all supported. Lossless and RGB coding. A common toolset is used for both lossy and lossless coding. RGB coding is supported either via the YCoCg integer color transform for maximum compression efficiency, or by

43、directly compressing RGB signals. Wavelet filters. A range of wavelet filters can be used to trade off performance against complexity. The Daubechies (9,7) filter is supported for compatibility with JPEG2000. A Fidelity filter is provided for improved resolution scalability. SMPTE ST 2042-1:2012 Pag

44、e 9 of 137 pages Simple stream navigation. The encoded stream forms a doubly-linked list with each picture header indicating an offset to the previous and next picture, to support field-accurate high-speed navigation with no parsing or decoding required. Multiple Profiles. VC-2 provides multiple pro

45、files to address the specific requirements of particular applications. Different profiles include or omit particular coding tools in order to best match the requirements of their intended applications. The Main profile provides maximum compression efficiency, variable bit rate coding and lossless co

46、ding using the core syntax. The Simple profile provides a less complex codec, but with lower compression efficiency, by using simple variable length codes for entropy coding rather than the arithmetic coding used by the Main profile. The Low Delay profile uses a modified syntax for applications requ

47、iring very low, fixed, latency. This can be as low as a few lines of input or output video. The Low Delay profile is suitable for light compression for the re-use of low bandwidth infrastructure, for example carrying HDTV over SD-SDI links. The High Quality profile similarly provides light compressi

48、on with low latency and also supports variable bit rate and lossless coding. SMPTE ST 2042-1:2012 Page 10 of 137 pages 1 Scope This standard defines the VC-2 video compression system through the stream syntax, entropy coding, coefficient unpacking process and picture decoding process. The decoder operations are defined by means of a mixture of pseudo-code and mathematical operations. VC-2 is an intra frame video codec that uses wavelet transforms together with entropy coding that can be readily implemented in hardware or software at very high bit rates. Additional standards and r