'· l
`
`Anlage D 5
`
`i.Sa. Google I VSL
`Nichtigkeitsklage vom 8. April 2015
`Bundespatentgericht Munchen
`Quinn Emanuel LLP
`
`ITU- Telecommunications Standardization Sector
`STUDY GROUP 16
`Video Coding Experts Group (VCEG)
`
`Document VCEG-N83 d1
`Filename: VCEG-N83d1.doc
`Generated: 21 Dec 2001
`
`Question: Q.6/SG16{VCEG)
`Source:
`Thomas Wiegand
`Heinrich Hertz Institute (HHI)
`Einsteinufer 37
`D-1 0587 Berlin, Germany
`H.26L Test Model Long-Term Number 9 (TML-9) draftO.
`
`Title:
`
`+49- (0)30 - 31002 617
`Tel:
`+49 - (0)30 - 392 72 00
`Fax:
`Email: wiegand@hhi.de
`
`Purpose:
`
`Test model
`
`This dO version of TML-9 includes the following changes from TML-8
`{as adopted cir announced in Santa Barbara):
`• Leaky bucket measurement [VCEG-N58r1]
`• Fast loop-filter algorithm [VCEG-N08]
`• Advanced Error Concealment [VCEG-N63]
`• Filtering outside of the picture [VCEG-N19]
`• Fast sub-pel interpolation method [VCEG-N31]
`• Robust Lagrangian Coder Control [VCEG-N38,VCEG-N39,VCEG-N50]
`• Supplemental enhancement information [VCEG-N60]
`• Clarification of B-picture coding
`• SP-frame Specification
`• Generalized ERPS syntax [VCEG-01 0]
`
`File:++VCEG-N83dl.doc
`
`Page: 1
`
`Vedanti Systems Limited - Ex. 2006
`Page 1
`
`

`
`E' t:.": + o vcg:;. Nil:Jd: . d O<":
`
`Pag~ : 2
`
`Vedanti Systems Limited - Ex. 2006
`Page 2
`
`

`
`CONTENTS
`
`3.5
`
`.................................................................................................................................................. 6
`I Scope
`2 Organization of source- and compressed data ......................................................................................... 7
`2.1
`Picture formats ......................................................................................................................... 7
`2.2 Subdivision of a picture into macro blocks ............................................................................... 7
`2.3 Order of the bitstream within a macroblock ............................................................................. 7
`2.4
`Syntax ....................................................................................................................................... 9
`2.4.1 Syntax diagram ..... .. ....................................................................................................... 9
`3 Description of syntax elements .............................................................................................................. 10
`3.1
`Picture sync ............................................................................................................................ 10
`3.2 Picture type (Ptype) ................................................................................................................ 10
`3.3 RUN .................................... , .................................................................................................. 10
`3.4 Macro block type (MB_Type) ................................................................................................ 10
`3.4.1 Intra
`...................................................................................................................... I 1
`3.4.2 Inter
`.................................................. .................................................................... 11
`Intra prediction mode (lntra_pred_mode) .............................................................................. 11
`3.5 .1 Mode 0: DC prediction ................................................. ................................................ 11
`3.5.2 Mode 1: Vertical/Diagonal Prediction ............... .......................................................... 11
`3.5.3 Mode 2: Vertical prediction ..................... .. .................................................... .............. 12
`3.5.4 Mode 3: Diagonal prediction ......................................................................... .............. 12
`3.5.5 Mode 4: Horizontal prediction ..................................................................................... 12
`3.5.6 Mode 5: Horizontal/Diagonal prediction ..................................................... ................ 12
`3.5.7 Prediction of chroma blocks .............................................................. ........................... 12
`3.5.8 Coding of Intra prediction modes ..................................................... .. ......................... 13
`3.5.9 Intra mode based on 16x16 macroblocks (16xl6 intra mode) .......... ........... ................ 14
`3.6 Reference frame (Ref_frame) ............................................................................... .................. 15
`3.7 Motion Vector Data (MVD) ....................................... ................................. ........................... 15
`3.7 .1 Fractional pixel accuracy ................................................................. ............................ 15
`3. 7.2 Prediction of vector components ................... .. ................................ ............................. 17
`3.7.3 Chroma vectors ............................................................................................................ 18
`3.8 Coded Block Pattern (CBP) .................................................................................................... 18
`3.9 Dquant .................................................................................................................................... 19
`4 Transform and inverse transform ........................................................................................................... 19
`4.1
`4x4 block size ......................................................................................................................... 19
`4.2
`2x2 transform/inverse transform of chroma DC coefficients ................................................. 19
`4.3 Zig-zag Scanning and quantization ........................................................................................ 20
`4.3.1 Simple zig-zag scan ...................................................................................................... 20
`4.3.2 Double zig-zag scan ........................ ~ ............................................................................ 20
`4.3.3 Quantization ................................................................................................................. 20
`4.3.4 Scanning and quantization of 2x2 chroma DC coefficients ......................................... 21
`4.4 Use of 2-dimensional model for coefficient coding ............................................................... 21
`4.5 Deblocking Filter .................................................................................................................... 21
`4.5.1 Picture Content Dependant Parameters for Filtering .............................. ..................... 22
`4.5.2 Filtering Process ................................................................. .......................................... 22
`4.5.3 Stronger filtering for intra coded macro blocks ................. ........................................... 23
`5 Entropy Coding ...................................................................................................................................... 24
`5 .1 Universal Variable Length Coding (UVLC) .......................................................................... 24
`5.2 Context-based Adaptive Binary Arithmetic Coding (CABAC) ............................................. 26
`5.2.1 Overview ...................................................................................................................... 26
`5.2.2 Context Modeling for Coding of Motion and Mode Information ................................ 26
`5.2.3 Context Modeling for Coding of Texture Information ................................................ 29
`5.2.4 Binarization of Non-Binary Valued Symbols .............................................................. 30
`5.2.5 Adaptive Binary Arithmetic Coding ............................................................................ 31
`6 Data Partitioning and Interim File Format.. ................................................................... ........................ 32
`
`File:++VCEG-N83dl.doc
`
`Page: 3
`
`Vedanti Systems Limited - Ex. 2006
`Page 3
`
`

`
`6.1 Data Partitioning ..................................................................................................................... 32
`7 Multi-Frame Buffering Syntax .............................................................................................................. 33
`7.1 Reference Picture Selection Flags (RPSF) ............................................................................. 33
`7.2 Picture Number (PN) ................................................................. , ............................................ 33
`7.3 Reference Picture Selection Layer (RPSL) ............................................................................ 33
`7.4 Re-Mapping of Picture Numbers Indicator (RMPNI) ............................................................ 33
`7.4.1 Absolute Difference of Picture Numbers (ADPN) ...................................................... 34
`7.4.2 Long-term Picture Index for Re-Mapping (LPIR) ....................................................... 35
`7.5 Reference Picture Buffering Type (RPBT) ............................................................................ 35
`7.5.1 Memory Management Control Operation (MMCO) .................................................... 36
`7.6 Multi-Picture Decoder Process ............................................................................................... 37
`7.6.1 Decoder Process for Short/Long-term Picture Management ....................................... 38
`7.6.2 Decoder Process for Reference Picture Buffer Mapping ............................................. 38
`7.6.3 Decoder Process for Multi-Picture Motion Compensation .......................... .. .............. 39
`7.6.4 Decoder Process for Reference Picture Buffering ....................................... ................ 39
`8 B-pictures ............................................................................................................................................... 41
`8.1
`Introduction ............................................................................................................................ 41
`8.2
`Five Prediction modes ............................................................................................................ 41
`8.3
`Finding optimum prediction mode ......................................................................................... 42
`8.4
`Syntax .............................................................................................. , ...................................... 42
`8.4.1 Picture type (Ptype) and RUN ..................................................................... .. ............... 43
`8.4.2 Macro block type (MB_type) ....................................................................................... 43
`8.4.3 Intra prediction mode (Intra_pred_mode) .................................................................... 44
`8.4.4 Reference Frame (Ref_frame) ...................................................................................... 44
`8.4.5 Block Size (Blk_size) ...................................................................... ............................. 45
`8.4.6 Motion vector data (MVDFW, MVDBW) ................................................................... 45
`8.5 Decoder Process for motion vector ...................................... .......... ........................................ 45
`8.5.1 Differential motion vectors ................................................... ....................................... 45
`8.5.2 Motion vectors in direct mode ....................... .............................................................. 45
`
`9 SPt~tur~~~;~d~~~i~~·:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::;
`
`9.2 Syntax changes ....................................................................................................................... 47
`9.3
`SP-frame decoding ............... .................................................................................................. 47
`10 Hypothetical Reference Decoder .................. ......................................................................................... 49
`10.1 Purpose ............................... .................................................................................................... 49
`10.2 Operation of the HRD ............................................................................................................ 49
`10.3 Decoding Time of a Picture .................................................................................................... 49
`10.4 Schedule of a Bit Stream ........................................................................................................ 49
`10.5 Containment in a Leaky Bucket ............................................................................................. 49
`10.6 Bit Stream Syntax ................................................................................................................... 50
`10.7 Minimum Buffer Size and Minimum Peak Rate .................................................................... 50
`1 0. 8 Encoder Considerations (informative) ................................................................................... 51
`11 Supplemental Enhancement Information ............................................................................................... 53
`11.1 Syntax ...................................................................................................................................... 53
`Non-normative Encoder Recommendation .................... .. ............................. .... ........... 54
`Appendix I
`I.1 Motion Estimation and Mode Decision .................................................................................. 54
`I. 1.1 Low-complexity mode ............................... .................................................................. 54
`Ll.2 High-complexity mode .............................................................. ................................... 56
`I.2 Quantization ..................................................................................... ...................................... 59
`Elimination of single coefficients in inter macro blocks ......................................................... 59
`I.3
`1.3.1 Luma
`............................................................................................... , ...................... 59
`L3.2 ·Chroma
`...................................................................................................................... 60
`Encoding with Anticipation of Slice Losses ........................................................................... 60
`1.4
`Network Adaptation Layer for IP networks ................................................................. 61
`Appendix II
`ILl Assumptions ........................................................................................................................... 61
`
`File:++VCEG-N83dl.doc
`
`Page: 4
`
`Vedanti Systems Limited - Ex. 2006
`Page 4
`
`

`
`11.2 Combining of Partitions according to Priorities ..................................................................... 61
`11.3 Packet Structure ...................................................................................................................... 62
`II.4 Packetization Process ............................................................................................................. 62
`II.S De-packetization ..................................................................................................................... 62
`II.6 Repair and Error Concealment ..................................................................................... .......... 63
`Appendix III
`Interim File Format ...................................................................................................... 64
`III. I General ................................................................................................................................... 64
`III.2 File Identification ................................................................................................................... 64
`III.3 Clump ..................................................................................................................................... 64
`III.3.1Definition ............................... ...................................................................................... 64
`III.4 Clump Order ...................................... ..................................................................................... 65
`IlLS Clump Definitions .................................................................................................................. 65
`III.S.l File Type Clump .......................................................................................................... 65
`III.5.2File Header Clump .................. ............................... .. .................................................... 66
`III.5.3Content Info Clump ...................................................................................................... 67
`III.5.4Altemate Track Info Clump ......................................................................................... 68
`III.5.5Parameter Set Clump ............................................................................... ........ ............. 69
`III.S.6Segment Clump ................................. ...................................................... ..................... 71
`III.S.7 Alternate Track Header Clump .................................................................................... 71
`III.5.8Altemate Track Media Clump ..................... .................... .......... ....................... ........... 73
`III.5.9Switch Picture Clump .................................................................................................. 73
`Appendix IV
`Non-Normative Error Concealment.. ........................................................................... 75
`IV.l Introduction ............................................................................................................................ 75
`IV.2 INTRA Frame Concealment. ............................................ ................................ ...................... 75
`IV.3 INTER and SPFrame Concealment .......................................... .................... ............ ............. 76
`IV.3.1 General
`...................................................................................................................... 76
`IV.3.2Concealment using motion vector prediction ............................................................... 77
`IV.3.3Handling of Multiple reference frames ........................................................................ 78
`IV.4 B Frame Concealment .................................................................................................... ; ....... 78
`IV.S Handling of Entire Frame Losses ........................ ................................................................... 78
`
`'ile:++VCEG- N83dl.doc
`
`Page: 5
`
`Vedanti Systems Limited - Ex. 2006
`Page 5
`
`

`
`1 Scope
`This document is a description of a reference coding method to be used for the development of a new
`compression method ITU-T recommendation- H.26L. The basic configuration of the algorithm is similar
`to H.263.
`Some of the diffe.rences from H.263 are:
`• Only one regular VLC or context-based adaptive binary arithmetic coding is used for symbol
`coding
`'A-sample or 1/8-sample accuracy used for motion prediction
`•
`• A number of different block sizes are used for motion prediction
`• Residual coding is based on 4x4 blocks and a integer transform is used
`• Multiple reference frames may be used for prediction
`The first part of the document describes the coding method by mainly defining the decoder actions.
`Towards the end, 'test model issues' relevant for the encoder to be used as reference for the development
`of the standard will be covered. This split should make it easy at a later stage to split the document into
`what is relevant for the standard and a test model. However, at the moment we find it preferable to have
`one combined document.
`
`Fi l e:++ VC~G -N 83dl.d o c
`
`Page: 6
`
`Vedanti Systems Limited - Ex. 2006
`Page 6
`
`

`
`32 Organization of source- and compressed data
`
`3-dl..lPicture fonnats
`At the moment only the QCIF and CIF formats are included in the model [Not true of the software
`anymore - the document needs to be updated about this (but not relevant to testing at QCIF and CIF
`resolutions)].
`
`~2..2.Subdivision of a picture into macroblocks
`A CIF picture is divided into 18x22 = 396 macroblocks. Similarly a QCIF picture is divided into 99
`macro blocks as indicated in FIGURE 1. At the moment there are no other layers in the described model.
`[Slice layers are supported in the software, but not fully reflected in this document.]
`
`QCIFimage
`
`.....
`...
`__...,
`
`9
`
`11
`
`FIGURE 1
`
`Subdivision of a QCIF picture into 16x16 macroblocks
`
`3.32.i0rder of the bitstream within a macroblock
`FIGURE 2 and FIGURE 3 indicate how a macroblock is divided and the order of the different syntax
`elements resulting from coding a macro block.
`
`Mode 1
`
`Mode 2
`
`Mode 3
`
`Mode 4
`
`ModeS
`
`~
`l±11J
`
`5
`7
`
`Mode7
`0 1 2 3
`4 5 6 7
`8 9 10 11
`12 13 14 15
`
`[J[IJE8ffi
`Mode6 m 3
`
`eile:+ +VCEG-N93dl.doc
`
`Page: 7
`
`FIGURE2
`
`Vedanti Systems Limited - Ex. 2006
`Page 7
`
`

`
`,.
`
`Numbering of the vectors for the different blocks depending on the inter mode. For each block the
`horizontal component comes first followed by the vertical component
`
`CBPY 8x8 block order
`
`0
`
`2
`
`1
`
`3
`
`Luma residual coding 4x4 block order
`
`Chroma residual coding 4x4 block order
`
`0
`
`2
`
`8
`
`1
`
`3
`
`9
`
`10
`
`11
`
`4
`
`6
`
`12
`
`14
`
`5
`
`7
`
`13
`
`15
`
`u
`
`v
`
`0
`
`2x2DC
`
`23
`
`AC
`
`20
`
`21
`
`24
`
`25
`
`FIGURE3
`
`Ordering of blocks for CBPY and residual coding of 4x4 blocks
`
`File:++VCEG- N83dl . d oc
`
`Page: 8
`
`Vedanti Systems Limited - Ex. 2006
`Page 8
`
`

`
`..
`
`M2..4.Syntax
`
`3.4.!2.4.1 Syntax diagram
`
`Omit
`
`c=::J Loop
`
`I
`
`I
`
`I
`I
`
`I
`
`I
`
`.I
`-'W
`[ Picture sync: TR, PQP, EOS
`
`\JI
`
`~
`
`t
`
`[ Ptype
`
`[RUN
`
`[ MB_Type
`
`I [ lntra_pred_mode
`,..,.
`w
`
`[ Ref_frame
`
`..._
`
`I [ MVD
`
`[ CBP
`
`\);
`
`[ Dquant
`
`I [ Tcoeff_luma
`
`,..,.
`..._
`
`I [ Tcoeff_chroma_DC
`
`......
`I [ Tcoeff_chroma_AC
`,..,.
`..._
`
`l
`I
`
`I
`
`l
`
`I
`
`J
`
`J
`
`I
`
`J
`
`I
`
`I
`
`I
`
`I;
`
`FIGURE4
`
`Syntax diagram for all the elements in the video hitstream
`
`File:++VC8G-N83dl.doc
`
`Page: 9
`
`Vedanti Systems Limited - Ex. 2006
`Page 9
`
`

`
`43 Description of syntax elements
`
`PQP(S bits)
`
`4dllPicture sync
`The first codeword in a picture is 31 bits long (L = 31). It works as a picture sync, but it also contains an
`INFO part that has 15 bits. These bits are used for:
`Temporal reference. The value of TR is formed by incrementing its value in the
`TR(8 bits)
`temporally-previous reference picture header by one plus the number of skipped or
`non-reference pictures at the picture clock frequency since the previously transmitted
`one.
`Picture QP: Information about the quantizer QUANT to be used for luma for the
`picture. (See under Quantization concerning QUANT for chroma). The 5 bit
`representation is the natural binary representations of the values of QP which range
`from 0 to 31. QP is a pointer to the actual quantization parameter QUANT to be used.
`(See below under quantization). The range of quantization value is still about the
`same as for H.263, 1-31. An approximate relation between the QUANT in H.263 and
`QP is: QUANTH.263(QP) "'QPO(QP) = 2QP/6 . QPO() will be used later for scaling
`purposes when selecting prediction modes.
`0 indicates QCIF, 1 indicates CIF
`Formats
`0 for picture header. 1 indicates End Of Sequence
`EOS
`The sketch below indicates where the different bits are located within INFO.
`
`8b: TR
`
`5b: PQP
`
`llb: Formats
`
`lb:EOS
`
`FIGURES
`
`Syntax diagram for the Picture header.
`
`~1.2.Picture type (Ptype)
`Code_number =0: Inter picture with prediction from the most recent decoded picture only.
`Code_number = 1: Inter picture with possibility of prediction from more than one previous decoded
`picture. For this mode information reference picture for prediction must be signalled
`for each macro block.
`Code_number =2: Intra picture.
`Code_number =3: B picture with prediction from the most recent previous decoded and subsequent
`decoded pictures only.
`Code_number =4: B picture with possibility of prediction from more than one previous decoded picture
`and subsequent decoded picture. When using this mode, information reference frame
`for prediction must be signalled for each macro block.
`
`4-.JllRUN
`A macroblock is called skipped if no information is sent. In that case the reconstruction of an inter
`macroblock is made by copying the collocated picture material from the last decoded frame.
`(See
`separate definition for B-pictures). RUN indicates the number of skipped macroblocks in an inter- orB(cid:173)
`picture before a coded picture. If the last MB of the frame is not coded an additional codeword is added
`which points to a MB outside the picture.
`
`4.43.4.Macro block type (MB_Type)
`Refer to TABLE 5. There are different MB-Type tables for Intra and Inter frames.
`
`File:++VCEG-N83dl.doc
`
`Page: 10
`
`Vedanti Systems Limited - Ex. 2006
`Page 10
`
`

`
`-"'r
`
`Intra
`
`4Ad3.4.1
`Intra 4x4
`Imode, nc, AC
`
`Intra coding as defined in sections 0 to 3.5.8.
`See definition in section 3.5.9.3. These modes refer to 16xl6 intra coding.
`
`NxM (eg. 8x4)
`
`No further information about the macro block is transmitted. A copy of the colocated
`macro block in the most recent decoded picture is used as reconstruction for the present
`macro block.
`The macro block is predicted from a past picture with block size NxM. For each NxM
`block motion vector data is provided. Depending on Nand M there may be 1 to 16
`sets of motion vector data for a macro block.
`4x4 intra coding.
`Intra4x4
`Code numbers from 9 and upwards represent l6x16 intra coding.
`
`4.S3 .. 5.Intra prediction mode (lntra_pred_mode)
`Even in Intra mode, prediction is always used for each sub block in a macroblock. A 4x4 block is to be
`coded (pixels labeled a to p below). The pixels A to I from neighboring blocks are already decoded and
`may be used for prediction.
`
`I F. B C D
`E a b c d
`F e f g h
`G i
`j k
`l
`H m n o p
`
`FIGURE6
`
`Syntax diagram for the Picture header.
`
`There are 6 intra prediction modes labeled 0 to 5. Mode 0 is 'DC-prediction' (see below). The other
`modes represent directions of predictions as indicated below.
`
`2
`
`3
`
`4
`
`5
`
`FIGURE 7
`
`Syntax diagram for the Picture header.
`
`4.&13.5.1 Mode 0: DC prediction
`Generally all pixels are predicted by (A+B+C+D+E+F+G+H)//8. If four of the pixels are outside the
`picture, the average of the remaining four is used for prediction. If all 8 pixels are outside the picture the
`prediction for all pixels in the block is 128. A block may therefore always be predicted in this mode.
`
`4-S.23.5.2 Mode 1: Vertical/Diagonal Prediction
`This mode is used only if all A,B,C,D are inside the picture.
`a is predicted by:
`(A+B)/2
`
`File:++VCEG-N83dl.doc
`
`Page: 11
`
`Vedanti Systems Limited - Ex. 2006
`Page 11
`
`

`
`B
`e is predicted by
`(B+C)/2
`b,i are predicted by
`c
`f,m are predicted by
`(C+D)/2
`c,j are predicted by
`d,g,h,k,l,n,o,p are predicted by D
`
`4.3.33.5.3 Mode 2: Vertical prediction
`If A,B,C,D are inside the picture, a,e,i,m are predicted by A, b,f,j,n by B etc.
`
`4£43.5.4 Mode 3: Diagonal prediction
`This mode is used only if all A,B,C,D,E,F,G,H,I are inside the picture. This is a 'diagonal' prediction.
`m is predicted by:
`(H+2G+F)//4
`i,n are predicted by
`(G+2F+E)/f4
`e,j,o are predicted by
`(F+2E+D//4
`(E+2I+A)//4
`a,f,k,p are predicted by
`b,g,l are predicted by
`(I+2A+B)//4
`c,h are predicted by
`(A+2B+C)//4
`(B+2C+D)//4
`d is predicted by
`
`4.-S.S3.5.5 Mode 4: Horizontal prediction
`If E,F,G ,H are inside the picture, a,b,c,d are predicted by E, e,f,g,h by F etc.
`
`4-.5.63.5.6 Mode 5: Horizontal/Diagonal prediction
`This mode is used only if all E,F,G,H are inside the picture.
`a is predicted by:
`(E+F)/2
`b is predicted by
`F
`c,e are predicted by
`(F+G)/2
`G
`f,d are predicted by
`i,g are predicted by
`(G+H)/2
`h,j,k,l,m,n,o,p are predicted by H
`
`4£.'7-3.5.7 Prediction of chroma blocks
`For chroma prediction there is only one mode. No information is therefore needed to be transmitted. The
`prediction is indicated in the figure below. The 8x8 chroma block consists of 4 4x4 blocks A,B,C,D.
`SO, 1 ,2,3 are the sums of 4 neighbouring pixels.
`If SO, Sl, S2, S3 are all inside the frame:
`A= (SO+ S2 + 4)/8
`B = (Sl +2)/4
`C = (S3 + 2)/4
`D=(Sl+S3+4)/8
`
`If only SO and Sl are inside the frame:
`A= (SO+ 2)/4
`B = (Sl +2)/4
`C =(SO+ 2)/4
`D = (Sl + 2)/4
`
`E"ile: ++VCEG-N83dl.doc
`
`Page: 12
`
`Vedanti Systems Limited - Ex. 2006
`Page 12
`
`

`
`.•
`
`If only S2 and S3 are inside the frame:
`A= (S2 + 2)/4
`B = (S2 + 2)/4
`C = (S3 + 2)/4
`D = (S3 + 2)/4
`If SO, S 1, S2, S3 are all outside the frame: A = B = C = D = 128
`(Note: This prediction should be considered changed)
`so
`
`Sl
`
`S2
`
`S3
`
`A
`
`c
`
`B
`
`D
`
`4£83.5.8 Coding of Intra prediction modes
`Since each of the 4x4 luma blocks shall be assigned a prediction mode, this will require a considerable
`number of bits if coded directly. We have therefore tried to find more efficient ways of coding mode
`information. First of all we observe that the chosen prediction of a block is highly correlated with the
`prediction modes of adjacent blocks. This is illustrated in FIGURE 8a. When the prediction modes of A
`and B are known (including the case that A or B or both are outside the picture) an ordering of the most
`probable, next most probable etc. of C is given. This ordering is listed in
`TABLE 1. For each prediction mode of A and B a list of 5 numbers is given. Example: Prediction mode
`for A and B is 2. The string 2 1 0 3 4 5 indicates that mode 2 is also the most probable mode for block C.
`Mode 1 is the next most probable one etc. In the bitstream there will for instance be information that
`ProbO = 1 (see TABLE 5) indicating that the next most probable mode shall be used for block C. In our
`example this means Intra prediction mode 1. Use of '-' in the table indicates that this instance can not
`occur because A or B or both are outside the picture.
`For more efficient coding, information on intra prediction of two 4x4 luma blocks are coded in one
`codeword (ProbO and Probl in TABLE 5). The order of the resulting 8 codewords is indicated in
`FIGURE 8b.
`
`2
`
`3
`
`6
`
`7
`
`0
`
`1
`
`4
`
`5
`
`2
`
`3
`
`6
`
`7
`
`0
`
`1
`
`4
`
`5
`
`b
`
`FIGURES
`
`a
`
`a) Prediction mode of block C shall be established. A and Bare adjacent blocks. b) order of intra
`prediction information in the bitstream
`
`TABLE 1
`
`Prediction mode as a function of ordering signalled in the bitstream (see text)
`outside
`4
`2
`
`5
`
`0
`
`B\ ?..
`
`Fi l e:++VCEG- N83 dl.doc
`
`Page: 13
`
`Vedanti Systems Limited - Ex. 2006
`Page 13
`
`

`
`102--- 201-- 012--- 012--- 012---
`021---
`outside 0-----
`041352 104325 230415 304215 043152 043512
`0
`045---
`014325 102435 203145 032145 041325 014352
`1
`045---
`012345 102345 210345 302145 042135 013245
`2
`045---
`304152 310425 231054 304215 403512 305412
`3
`045---
`403512 401532 240351 430512 403512 405312
`4
`405---
`504--- 540312 015432 201453 530412 450312 504132
`5
`
`Intra mode based on 16xl6 macroblocks (16x16 intra mode)
`4.&93.5.9
`This intra mode is particularly suitable for regions with little details, also referred to as 'flat' regions.
`
`•
`
`•
`
`Prediction modes
`4£9...l3.5.9.1
`Assume that the block to be predicted has pixel locations 0 to 15 horizontally and 0 to 15 vertically. We
`use the notation P(i,j) where i,j = 0 .. 15. P(i,-1), i=0 .. 15 are the neighboring pixels above the block and P((cid:173)
`l,j), j=0 .. 15 are the neighboring pixels to the left of the block. Pred(i,j) i,j = 0 .. 15 is the prediction for the
`whole luma macroblock. We have 4 different prediction modes:
`IMODE = 0 (vertical)
`•
`Pred(i,j) = P(i,-1), i,j=0 ..