SMPTE ST 329M-2000 Television - MPEG-2 Video Recoding Data Set - Compressed Stream Format.pdf

1、STD=SMPTE 329M-ENGL 2000 8357401 OOOLiBSb 185 SMPTE STANDARD SMPTE 329M-2000 Table of contents 1 Scope 2 Normative references 3 General 4 Definitions 5 Compressed stream format of the MPEGP Annex A Bibliography recoding data set 1 Scope This standard specifies the stream format of the MPEG-2 recodin

2、g data set for the representation of compressed ISO/IEC 13818-2 MPEG coding informa- tion, as used in applications requiring transport systems of reduced data capacity. The coding information is derived from an ISOAEC 1381 8-2 compliant MPEG bit stream during the decoding process, as described in IS

3、OAEC 13818-2. The information based on this stream format may be transported by various means; for example, the elementary stream format defined in SMPTE 328M. 2 Normative references The following standards contain provisions which, through reference in this text, constitute provisions of this stand

4、ard. At the time of publication, the editions indicated were valid. All standards are subject to revision, and parties to agreements based on this standard are encouraged to investigate the possibility of applying the most recent edition of the standards indicated below. Paga 1 of 14 pages SMPTE 327

5、M-2000, Television - MPEG-2 Video Recoding Data Set ITU-T H.262, ISOAEC 1381 8-2:1996, Information Technology - Generic Coding of Moving Pictures and Associated Audio Information: Video 3 General The main application of this standard is to preserve the quality of the video signal when cascading deco

6、ders and coders for editing or transcoding purposes by feeding forward previous coding decisions. The MPEG-2 recoding data set in combination with an MPEG-2 decoded or partially decoded picture is eff ectivefor implementing editing or transcoding of the MPEG-2 bit stream. There are, however, applica

7、tions in which the transmission of all the recoding data set is not possible. Some current equipment may have restricted capacity for the transmission of the recoding data. This limitation has an impact on subsequent compression stages which can make use of the MPEG-2 recoding process. In order to d

8、ecrease the bfi rate for the recoding data set, the MPEG-2 recoding data set is converted into an MPEG-like stream, which is called the compressed stream format of the MPEG-2 recoding set. This standard defines this stream format. The compressed stream format of the MPEG-2 recod- ing set much resemb

9、les an MPEG-2 video stream, except that the compressed stream format of the MPEG-2 recoding set does not have DCT coeff icients which occupy the major part of an MPEG video stream. Copyright 02ooo by THE SOCIEPI OF MOTION PICTURE AND TELEVISION ENGINEERS 595 W. Hartsdak Ave., white Plams, NY 10607 (

10、914) 761-llW Approved January 20,2000 STD-SMPTE 329M-ENGL 2000 II 8357401 0009057 O11 m By adopting an MPEG-like form for the compressed stream format of the MPEG-2 recoding set, use is made of an efficient compression for this data set for storage and transmission operations. Moreover, it may be po

11、ssible to reuse the MPEG decoding devices to decode this compressed stream format of the MPEG-2 recoding set. The compressed format of the MPEG-2 recoding set syntactical structure is modified by red-bw-flag and red-bw-indicators. The selection of one of the vari- ations of the syntactical structure

12、 will be determined by the application by taking into account a balance between bit rate and the number of elements in the recoding data set. This standard does not describe the transport mecha- nism, but describes the stream structure of the compressed stream format of the MPEG-2 recoding set. This

13、 stream format is independent of application, and all the transport information in the reduced band- width recoding data transportation system should be based on this stream. The transport mechanism depends on application, which should be defined in other standards docu- ments. The first of these is

14、 the MPEG-2 elementary stream editing information, SMPTE 328M. 4 Definitions 4.1 btstream: An ordered series of bits con- forming to ISOAEC 1381 8-2. 4.2 cascading: This describes the process where video that has once been coded (com- pressed) is subsequently decoded and coded once more. This cascad

15、ing step could carry on for any number of generations. 4.3 coding: This is the process by which an un- compressed video sequence is compressed to a bit stream that conforms to the ISOAEC 13818-2 standard. 4.4 compressed stream format of the MPEG-2 recoding data set: This is the compressed infor- mat

16、ion of the recoding data set which is used in reduced bandwidth recoding data transportation systems. 4.5 decoder: A compressed bit stream decoder that complies with the ISOAEC 13818-2 standard. 4.6 macroblock Defined in the ISO/IEC 1381 8-2 standard as a block of 16 x 16 luminance pixels. 4.7 MPEG

17、profileievel: As defined in ISO/IEC 1381 8-2. 4.8 picture: As defined in ISOAEC 13818-2. 4.9 recoding data set: The set of information de- rived from an ISOAEC 1381 8-2 compliant stream which, when accompanied by decoded or par- tially decoded video information, assists in the recreation of the orig

18、inal MPEG-2 stream. 4.10 reduced bandwidth recoding data transpor- tation system: This is the system that does not have enough bandwidth capacity for transport- ing a full set of the recoding data set, but is able to transport the compressed stream format of the MPEG-2 recoding data set. 4.1 1 reduc

19、ed bandwidth stream: T h e h i g h est syntactic structure of the reduced MPEG-2 recoding data set. It contains the recoding infor- mation for one picture. 4.12 stripe: A horizontal row of macroblocks spanning the full width of the decoded picture (stripes are numbered from top to bottom start- ing

20、at zero). 4.13 transcoding: A conversion within the MPEG-2 stream domain, such as bit rate chang- ing or changing the group of pictures (GOP) structure. 5 Compressed stream format of the MPEG-2 recoding data set 5.1 Compressed stream format of the MPEG-2 recoding set syntax The compressed stream for

21、mat of the MPEG-2 re- coding set is a subset of the ISOAEC 1381 8-2 stream and is extracted from the original MPEG-2 video stream. 5.1.1. Start codes Start codes are specific bit patterns that do not other- wise occur in the compressed stream format of the MPEG-2 recoding data set. Page 2 of 14 page

22、s STD=SMPTE 329M-ENGL 2000 8357qO1 0004058 T58 Each start code consists of astart code prefix followed by a start code value. The start code prefK is a string of 23 bits with the value zero followed by a single bit with the value one. The start code value is an eight-bit integer that iden- tifies th

23、e type of start code. Most types of start code have just one start code value. However, slice-startcode is represented by many start code values. In this case, the start code value is the slice-verticalgosition for the slice. All start codes shall be byte aligned. When necessary, this shall be achie

24、ved by inserting bits with the value zero before the start code prefix such that the first bit of the start code prefix is the first (most significant) bit of a byte. Table 1 defines the start de values for the start codes used in the compressed stream format of MPEGP recoding set. The use of the st

25、art des is defined in the following syntax description with the exception of the sequence-error-code. The sequence-error-code has been allocated for use by a media interface to indicate where uncorrectable errors have been de- tected (see tables 2 - 22). 5.2 Compressed stream format of the MPEG-2 re

26、coding set semantics Almost all of the syntax elements in the compressed stream format of the MPEG-2 recoding set have the same meaning as those defined in ISOAEC 13818-2. Some elements are newly defined and some have a different meaning. Definitions for these elements are given below. 5.2.1 Recodin

27、g stream information Re-coding-stream-info-ID: A 16-bit integer that is used for identification of re-coding-stream-info(). The value is 1001 o001 1110 1100 (O x 91 ec). red-bw-flag: A 1-bit flag. Its definition is given in SMPTE 327M. red-bw-indicator: A 2-bit integer defined in SMPTE 327M. num-oth

28、er-bits: A 7-bit integer. This syntax element is defined in SMPTE 327M. SMPTE 329M-2 num-mv-bits: An 8-bit integer defined in SMPTE 327M. num-coef-bits: A 14-bit integer. This syntax element is defined in SMPTE 327M. num-coef-bts, num-mv-bits, and num-other-bits exist when red-bw-flag is O. These th

29、ree data sets are repeated by the number of macroblocks. The data sets match the macroblocks respectively, which are scanned from left top to right bottom horizontally first, as in MPEG-2. The value of red-bw-indicator specifies the level of reduced information and it ranges from O to 3. As the valu

30、e is increased, the more information is reduced. As the compressed stream format of the MPEG-2 recoding data set syntax structure depends on this red-bw-flag and red-bw-indicator, the decoder shall recognize these elements in the current bit stream. Then it shall parse the bit stream accordingly. Th

31、e details of each red-bw-indicator value are de- scribed as follows: - Indicator O: This stream transports all the recod- ing data set, except that the information such as num-coef-bits, num-mv-bits, and num-other-bits are not transported. All the recoding data set couid be derived from this stream.

32、 Therefore, the position of slice-start, skipped-mb, and mb-quant shouid be the same as in the original stream. - Indicator 1 : This stream is the same as indicator O, except that codedblockjattem0 infomiation is not transported. This stream is not guaranteed to have the values of slice-start, skipp

33、ed-mb, and mb-quant to be the same as in the original stream. - Indicator 2: This stream transports only quan- tizer-scale-code information in slice() and macro- block(). Other information which is carried in slice() and macroblock() is not guaranteed to be the same as in the original stream. - Indi

34、cator 3: The infomiation of slice() and macro- block() is not transported by this stream. The relation between red-bw-indicator and the elements of the MPEG-2 recoding data set de- scribed in SMPTE 327M is shown in table 23. STD.SMPTE 329M-ENGL ZOCO 08357401 000057 974 R SMPTE 329Nt-m pict Ure-start

35、-code slice-start-code reserved Table 1 - Start code value _ O0 O1 through AF BO Name user-data-start-code sequence-headercode se q uence-erro r-Co de I I Start code value hexadecimal) 82 83 84 reserved sequence-end-code group-start-code I reserved I 81 I B6 B7 88 CO mp ressed-streamformat-of-M P EG

36、-2-re-coding-s next-start-code() I ext e ns ion-s t art-code I 85 I No. of bits Mnemonic extensions-and-user-data(2) if (! red-bwflag II (red-bw-indicator =2) Dict Ure-datal) Table 2 - Compressed stream format of the MPEG-2 recoding set cequence-end-code 1 32 bslbf I sequence-header() I I I _ sequen

37、ce-extension() extension-and-user-data(0) if (nextbitso = group-start-code) I group-of-pictures-header() I I I extension-and-user-data(1) 1 picture-header0 I picture-coding-extension() I I I I re-coding-stream-info() I I I Page 4 of 14 pages Table 3 - Sequence header sequence-header() I No. of bits

38、Mnemonic I horizontal-size-value t 12 I uimsbf I vertical-size-value aspect-ratio-information f rame-rate-code bitpratepvalue 12 uirnsbf 4 uimsbf 4 uimsbf 18 uimsbf I marker- b it I 1 I bslbf I vbv-buff er-size-value constrained-parameters-flag load-intra-quantizer-matrix if (load-intra-quantizer_ma

39、trix) intra-quantizer-matrix64 load-non-intra-quantizer-matrix if (load-non-intra-quantizer-matrix) next-start-code() non-intra-quantizer-matrix64 1 10 uimsbf 1 bslbf 1 uimsbf a+64 uimsbf 1 uirnsbf 864 uirnsbf Table 4 - Sequence extension Page 5 of 14 pages STDiSMPTE 327M-ENGL 2000 I8357403 000LiObl

40、 5Li2 SMF“E329M-2000 extension-and-user-data(i) while ( (nextbitso = extension-start-code) II (nextbitso = user-data-start-code) ) extension-data() user-data() if (i ! = 1) inumber-of-frame-center-offsets; i+) frame-center-horizontal-offset marke r-b it f rame-center-vertical-off set marker-bit 1 ne

41、xt-startcode() 1 Table 8 - Quant matrix extension No. of bits Mnemonic 4 uimsbf 16 simsbf 1 bslbf 16 simsbf 1 bslbf Table 10 - Copyright extension next-starLCode() I I 1 STD=SMPTE 327M-ENGL 2OUO 8357iOL 00040b3 315 = SMPTE 329M-2000 user-data() user-data-start-code while (nextbitso ! = O000 O000 O00

42、0 O000 O000 0001) user-data Table 11 - User data No. of bits Mnemonic 32 bslbf a uimsbf next-star-code() I Table 12 - Group of pictures header g ro u p-of-p ict u r es-h eade r() group-start-code time code No. of bits Mnemonic 32 bslbf 25 bslbf cios ed-g o p broken-lin k next-s t art-co de ( ) Table

43、 13 - Picture header 1 uirnsbf 1 uimsbf while (nextbitso = 1) I next-startcode() I 1 I extra-inf ormation-picture 1 extra-bit-picture /* with the value O */ Page 8 o 14 pages uimsbf 1 uirnsbf a STD-SMPTE 327M-ENGL 2000 I8357401 000LiOb4 251 m WPTE 329M-2Ooo Table 14 - Picture coding extension Table

44、15 - Recoding stream information next-statt-code () 1 I I STD*SMPTE 327fl-ENGL 2000 8357403 000LiObS 198 II SYPTE329M-20m picture-data() Table 16 - Picture data No. of bits Mnemonic slice() slice-start-code slice() 1 while (nextbits0 = slice-start-code) No. of bits Mnemonic 32 bslbf _ next-starLCode

45、() 1 quantizer-scale-code if (nextbits0 = 1) 4 Table 17 - Slice - 5 uimsbf intra-slicef lag 1 bslbf intra dice I 1 uimsbf reserved-bits while (nextbitso = 1) extrabit-slice /* with the value 1 */ 7 uimsbf 1 uimsbf do macroblock0 ext ra-inf or mat ion-slice while (nextbits() ! = O00 O000 O000 O000 O0

46、00 0000) next-st art-code i ) 8 uimsbf Page 10 of 14 pages 1 extra-bitslice /* with the value O */ 1 uimsbf STD-SMPTE 32il-ENGL 2000 W 8357q03 00040bb 024 m SMPTE 329M-2W Table 18 - Macroblock (! red-bwflag II coded block DatternO idicator = O) Table 19 - Macroblock modes Page 11 Of 14 pagea STD-SMP

47、TE 327M-ENGL 2000 m 8357qL 000110b7 Tb = if (motion-vector-count = 1) if (mvformat = field) & (dmv ! = 1) Table 20 - Motion vectors I motion_vectors is) I I No. of bits 1 Mnemonic I motion-verticalf ield-selectOs motion-vector(0,s) 1 else 4 1 uimsbf mot io n-ve rt ic alf i e Id-c e lect O SI motion-

48、vector(0,c) motion-verticalf ield-select 1 SI motion-vector(1 ,s) 1 1 uimcbf 1 uimsbf Table 21 - Motion vector coded-block-pattern() coded-block-pattern-420 if (chromaformat = 4:2:2) coded-bloc k-pattern-1 Table 22 - Coded block pattern No. of bits Mnemonic 3-9 vlclbf 2 uimsbf _ coded-block-pattern-

49、2 I if chroma format = 4:4:4) I I I 6 uimsbf Page 12 of 14 pages STD-SMPTE 329fl-ENGL 2000 E 8357401 0004Ob8 9T7 SMPTE 3291111-2000 1 1 O Table 23 - Reduced bandwidth indicators 1 1 1 1 1 1 11 1 1 O 1 X 1 xx X X O X X O xx 1 red-bw-indicator- 1 nu m-mv-bits, n um-ot h er-bits 1 -code, q-scalefype motion-type, mv_vertfieldslrlsl, mvrlsltl mb-rnfwd, mb-mbwd mb-pattern coded-block-pattern mb-int ra I siice-start Indicator O O 1 Indicator 1 O 1 Indicator