1、SMPTE 402M-2003 Page 1 of 7 pages 1 Scope This standard specifies a data reduction method for the MPEG-2 video recoding data set for SDTV as defined in SMPTE 327M. The coding information is as derived from an ISO/IEC 13818 compliant MPEG bit stream during the picture decoding process, as described i
2、n ISO/IEC 13818-2. This data-reduced MPEG-2 recoding data set may be transported by various means defined in other SMPTE standards (see annex for further details). 2 Normative references SMPTE 319M-2000, Television Transporting MPEG-2 Recoding Information through 4:2:2 Component Digital Interfaces S
3、MPTE 327M-2000, Television MPEG-2 Video Recoding Data Set 3 Definitions 3.1 cascading: The process where video that has once been coded (compressed) is subsequently decoded and coded once more. This cascaded step could carry on for any number of generations. 3.2 coding: The process by which an uncom
4、pressed video sequence is compressed to a bit stream that conforms to the ISO/IEC 13818-2 standard. 3.3 decoder: A compressed bit stream decoder that complies with the ISO/IEC13818-2 standard. 3.4 macroblock: The four 8 by 8 blocks of luminance data and the corresponding blocks of chrominance data d
5、epending on the MPEG-2 profile coming from a 16 by 16 section of the luminance component of the picture as defined in ISO-IEC 13818-2. 3.5 macroblock rate information: The coding information from the ISO/IEC 13818-2 bit stream that relates to the individual macroblocks as defined in this standard 3.
6、6 MPEG-2: The standard for the generic coding of moving pictures as defined in ISO/IEC 13818-2. 3.7 MPEG-2 video recoding data set: The data set for the representation of certain ISO/IEC 13818-2 MPEG coding information for the purpose of optimally cascading decoders and re-encoders as defined in SMP
7、TE 327M 3.8 picture: Source, coded, or reconstructed image data as defined in ISO/13818-2 Page 1 of 7 pagesSMPTE 402M-2003Copyright 2003 by THE SOCIETY OF MOTION PICTURE AND TELEVISION ENGINEERS 595 W. Hartsdale Ave., White Plains, NY 10607 (914) 761-1100 ApprovedSeptember 24, 2003SMPTE STANDARD for
8、 Television A Data Reduction Method for the MPEG-2 Video Recoding Data Set SMPTE 402M-2003 Page 2 of 7 pages 3.9 picture rate information: Coding information from the ISO/IEC 13818-2 bit stream that relates to the whole picture, with some additional information derived from sequence coded_block_patt
9、ern and coefficient information are removed from the bit stream; motion vectors are transmitted to one-quarter of their usual precision; the remaining bits of the motion vectors are transmitted at the end of the bit stream. The information required in the MPEG-2 video recoding data set can be broken
10、 down into two parts. These two parts consist of the picture rate information, described in clause 5, and the macroblock rate information, described in clause 6. SMPTE 327M defines a method for signaling reduced bandwidth operation in any particular transport mechanism using the MPEG-2 video recodin
11、g data set. This standard uses that method. In the case of the present specification, for data sets that have been recoded with data-reduction, the decoder shall assume that red_bw_flag is set to 1 and that red_bw_indicator is set to “Indicator 1”. This value of red_bw_indicator indicates that coded
12、_block_pattern, slice_start, mb_quant and skipped_mb are not transmitted. It will be noted that mb_quant and skipped macroblock information are in fact transmitted in the data-reduced bit stream, but this information is transmitted as part of the data reduction mechanism and does not automatically i
13、ndicate the values of those parameters in the original MPEG-2 bit stream. 5 Picture rate information The picture rate information consists of 4,320 bits, exactly as given in table 4 of SMPTE 319M. If the MPEG-2 picture_structure is “Field picture”, then the picture rate information for the two field
14、-pictures in the frame is transmitted in succession, a total of 8,640 bits. 6 Macroblock rate information The macroblock rate information is transmitted immediately following the picture-rate information. It consists of two parts. The first part is essentially the picture data in an MPEG-2 video ele
15、mentary stream without coefficients and containing reduced-precision motion vectors. The second part SMPTE 402M-2003 Page 3 of 7 pages consists of additional information which restores the precision of the motion vectors. The macro-block rate information is described in detail in the following table
16、s. 6.1 Syntax of macroblock rate information 6.1.1 Picture data This table is the same as 6.2.3.7 of ISO/IEC 13818-2, with the addition of motion_vector_precision_bits and its start-code. Note that if picture_structure is “Field picture”, then the slice() information is transmitted for the two field
17、-pictures in succession, followed by the motion_vector_precision_bits() for the two field-pictures in succession. picture_data() No. of bits Mnemonic do slice() while ( nextbits() = = slice_start_code ) next_start_code() motion_vector_precision_bits() 6.1.2 Slice This table is the same as 6.2.4 of I
18、SO/IEC 13818-2, except that slice_vertical_position_extension, scalable modes and extra_information_slice are not supported. In the data-reduced video recoding data set bit stream, each slice corresponds exactly to a row of macroblocks, regardless of the slice structure of the original MPEG-2 bit st
19、ream. Any whole bytes of slice stuffing present in the original MPEG-2 bit stream are removed. slice() No. of bits Mnemonic slice_start_code 32 bslbfquantiser_scale_code 5 uimsbf if( nextbits() = = 1) intra_slice_flag 1 bslbfintra_slice 1 uimsbf reserved_bits 7 uimsbf extra_bit_slice /* with the val
20、ue 0 */ 1 uimsbf do macroblock() while ( nextbits() != 000 0000 0000 0000 0000 0000) next_start_code() 6.1.3 Macroblock This table is the same as 6.2.5 of ISO/IEC 13818-2 except that coded_block_pattern() is not transmitted. macroblock() No. of bits Mnemonic while ( nextbits() = = 0000 0001 000 ) ma
21、croblock_escape 11 bslbfmacroblock_address_increment 1-11 vlclbf macroblock_modes() if( macroblock_quant ) SMPTE 402M-2003 Page 4 of 7 pages quantiser_scale_code 5 uimsbf if( macroblock_motion_forward | ( macroblock_intra if( f_codes0 1 ) lsb_expected1mbrs1 = TRUE; if( dmv = 1) dmvector0 motion_code
22、rs0 1-11 vlclbfif( f_codes1 1 ) if( f_codes1 1 ) lsb_expected1mbrs0 = TRUE; if(dmv = 1) dmvector1 6.1.4 Motion vector precision bits This table is specific to the data-reduced video recoding data set bitstream. It provides for the transmission of the second-least and least significant bits of the mo
23、tion vectors where those bits are recorded as being expected in the lsb_expected array. Note that there is no protection against start code emulation within the data, because no further start codes can occur within the current picture. motion_vector_precision_bits No. of bits Mnemonic motion_vector_
24、precision_bits_start_code /* takes the value 1011 0000 */ 8 uimsbffor( mb=0; mb high) vectorrst = vectorrst - range; if( ( mv_format = =”field” ) Note that lsb takes the value zero unless a value has been transmitted. SMPTE 402M-2003 Page 7 of 7 pages Annex A (informative) Purpose of this standard T
25、he aim of this data reduction method is to reduce the MPEG-2 video recoding data set data rate to below 1.5 Mb/s for standard definition systems so that the video recoding data set may be carried in an auxiliary transport path. If the data-reduced video recoding data set bit stream should exceed the
26、 available capacity of the transport mechanism for a given video frame, this bit stream is truncated. It is the responsibility of the re-encoder using this stream to take appropriate action in this eventuality. Examples of this path include an AES-3 audio channel and the auxiliary data packets of th
27、e 50-Mb/s structure for DV-based DIF streams. This standard exists because the data rate of SMPTE 327M (MPEG-2 video recoding data set) data exceeds the bandwidth of certain links on which it is desired to transport this information. The intention is to reduce the data rate of the SMPTE 327M data in
28、 such a way that the maximum quality benefit can be achieved from the data when transmitted on a restricted bandwidth channel such as the auxiliary data packets of the 50-Mb/s structure for DV-based DIF streams. Verification of the validity of the data carried by this data reduction standard is beyo
29、nd the scope of this standard. There are operational situations where picture information may change, but the related video recoding data carried by this standard remains unchanged. A simple example of this is where a particular picture transport system truncates the number of active lines without u
30、pdating the recoding information. Implementations which may encounter this situation should ensure steps are taken to check the integrity of the data where possible. Annex B (informative) Bibliography ISO/IEC 13818-2:2000, Information Technology Generic Coding of Moving Pictures and Associated Audio Information: Video
