SMPTE RP 2019-2-2016 VC-3 Decoder and Bitstream Conformance.pdf

1、 Copyright 2016 by THE SOCIETY OF MOTION PICTURE AND TELEVISION ENGINEERS 3 Barker Avenue, White Plains, NY 10601 (914) 761-1100 Approved May 18, 2016 Table of Contents Page Foreword . 2 Intellectual Property 2 1 Scope . 3 2 Conformance Notation . 3 3 Normative References . 3 4 Overview (Informative

2、) 4 5 Description of Testing Materials . 5 5.1 Reference Decoder Software . 5 5.1.1 Diagnostic Mode Output 5 5.1.2 Reference YCBCR Output . 6 5.2 Test Bitstreams and Data Files 9 6 Conformance Testing Procedure . 9 6.1 Decoder Conformance Requirements . 9 6.1.1 Decoder Front-End Conformance Testing

3、. 9 6.1.2 Inverse Discrete Cosine Transform Statistical Tests . 12 6.2 Bitstream Conformance Testing 14 Annex A Test Bitstreams . 15 A.1 Overview (Informative) 15 A.2 Bitstream List (Normative) 15 Annex B Bibliography (Informative) 24 Page 1 of 24 pages VC-3 Decoder and Bitstream Conformance SMPTE R

4、P 2019-2:2016 Revision of SMPTE RP 2019-2:2014 SMPTE RECOMMENDED PRACTICE SMPTE RP 2019-2:2016 Page 2 of 24 pages Foreword SMPTE (the Society of Motion Picture and Television Engineers) is an internationally-recognized standards developing organization. Headquartered and incorporated in the United S

5、tates of America, SMPTE has members in over 80 countries on six continents. SMPTEs Engineering Documents, including Standards, Recommended Practices and Engineering Guidelines, are prepared by SMPTEs Technology Committees. Participation in these Committees is open to all with a bona fide interest in

6、 their work. SMPTE cooperates closely with other standards-developing organizations, including ISO, IEC and ITU. SMPTE Engineering Documents are drafted in accordance with the rules given in its Standards Operations Manual. SMPTE RP 2019-2 was prepared by Technology Committee 10E. Intellectual Prope

7、rty SMPTE draws attention to the fact that it is claimed that compliance with this Standard may involve the use of one or more patents or other intellectual property rights (collectively, “IPR“). The Society takes no position concerning the evidence, validity, or scope of this IPR. Each holder of cl

8、aimed IPR has assured the Society that it is willing to License all IPR it owns, and any third party IPR it has the right to sublicense, that is essential to the implementation of this Standard to those (Members and non-Members alike) desiring to implement this Standard under reasonable terms and co

9、nditions, demonstrably free of discrimination. Each holder of claimed IPR has filed a statement to such effect with SMPTE. Information may be obtained from the Director, Standards or that a certain course of action is preferred but not necessarily required; or that (in the negative form) a certain p

10、ossibility or course of action is deprecated but not prohibited. The keywords “may“ and “need not“ indicate courses of action permissible within the limits of the document. The keyword “reserved” indicates a provision that is not defined at this time, shall not be used, and may be defined in the fut

11、ure. The keyword “forbidden” indicates “reserved” and in addition indicates that the provision will never be defined in the future. A conformant implementation according to this document is one that includes all mandatory provisions (“shall“) and, if implemented, all recommended provisions (“should“

12、) as described. A conformant implementation need not implement optional provisions (“may“) and need not implement them as described. 3 Normative References The following standards contain provisions which, through reference in this text, constitute provisions of this standard. At the time of publica

13、tion, 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. SMPTE ST 2019-1:2016, VC-3 Picture Compression and Data

14、 Stream Format SMPTE VC-3 Test Materials Reference Decoder Software and Bitstreams ISO/IEC 23002-1:2006, Information Technology MPEG Video Technologies Part 1: Accuracy Requirements for Implementation of Integer-Output 88 Inverse Discrete Cosine Transform SMPTE RP 2019-2:2016 Page 4 of 24 pages 4 Ov

15、erview (Informative) Figure 1 depicts the VC-3 decoder verification process diagram. There are two test sequences which the decoder will need to pass in order for a target decoder to successfully conform to SMPTE VC-3. The first test sequence is called Decoder Front-End Conformance Testing and is de

16、tailed in Section 6.1.1. The second test sequence is called the Inverse Discrete Cosine Transform Statistical Test. It is described in Section 6.1.2. Figure 1 Target Decoder Conformance Testing The purpose of the first test (Decoder Front-End Conformance Test) is to test the front-end of the VC-3 de

17、coding stages; i.e. the header parsing, entropy decoding, inverse quantization stages of the decoding process and decoding process of compressed alpha channel. This test compares the output of a reference decoders diagnostic output against the output of the target decoder. The reference decoder and

18、testing materials are described in Section 5. VC-3 test files and the software reference decoder are available at the Header Parsing Entropy Decoding Inverse Quantization Inverse Discrete Cosine Transform Decoder under test VLC Codeword Tables CompressQuantization Weighting Tables VC-3 RLE Compresse

19、d Alpha Input VC-3 Compressed Bitstream Decoder Front-End Conformance Test IDCT Precision Test Alpha Zig-Zag Difference Prediction RLE Decoding Size Control Binary Decoding Decoder RLE Alpha Test VC-3 RLE Compressed Alpha SMPTE RP 2019-2:2016 Page 5 of 24 pages SMPTE Store at http:/smpte.org. Naviga

20、te to the SMPTE Store, then to Test Materials, to locate the VC-3 materials and the corresponding license. The reference decoder conforms to the complete decoding process specified in SMPTE ST 2019-1. Since the Entropy Decoding, Inverse Quantization and RLE decoding stages are mathematically specifi

21、ed, a conforming decoders output from this processing chain will match exactly with another conforming decoders output. This enables the Decoder Front-End Conformance Test to be specified as a binary comparison between the reference decoders output and the target decoders output. The purpose of the

22、second test (Inverse Discrete Cosine Transform Statistical Test) is to measure the statistical properties of an Inverse Discrete Cosine Transform (IDCT) implementation. Since the implementation of the IDCT is not completely specified in SMPTE ST 2019-1, there can be small variations depending upon a

23、 decoders implementation. That is why the statistical measures listed in Section 6.1.2 are applied the precision and accuracy of a decoders IDCT are measured to verify that it meets the required statistical tolerance. The purpose of the third test is to ensure the lossless Alpha decoding works accor

24、ding to specification and that the decoder will return a result which can be used for binary comparison of the overall decoding results. The International Standards Organization (ISO/IEC) has written a document (ISO/IEC 23002-1) describing accuracy requirements for IDCT operations in video compressi

25、on systems. This Recommended Practice refers to the ISO/IEC documents when detailing the requirements of the Inverse Discrete Cosine Transform Statistical Tests. Note: The use of “YCBCR“ terminology in this document is consistent with its use in ST 2019-1 Finally, Section 6.2 describes the bitstream

26、 conformance testing procedure. 5 Description of Testing Materials 5.1 Reference Decoder Software The reference decoder which implements the complete decoding process specified in SMPTE ST 2019-1 shall be used for testing bitstream conformance. This sub-section describes the operation of the referen

27、ce decoder. The reference decoder, in the form of C+ source code and associated documentation, provides decoding functionality for VC-3 conformant bitstreams. It compiles under common “Windows, Linux and Mac OS X” operating systems. The source code will compile to a library, using a generic make fil

28、e. The reference decoder supports two modes of operation: “Diagnostic Output” and “Reference Channel Output”. These are described further in following sub-sections. 5.1.1 Diagnostic Mode Output The diagnostic mode output of the reference decoder shall be utilized to verify a candidate decoders confo

29、rmance to the VC-3 standard as described in Section 6.1.1. When operating in this mode, the reference decoder will perform the header parsing, the entropy decoding, and the inverse quantization processing steps as shown in Figure 2. SMPTE RP 2019-2:2016 Page 6 of 24 pages Figure 2 Reference Decoder

30、in Diagnostic Mode When the reference decoder is in the diagnostic mode, it will generate diagnostic output files containing RLE Compressed Alpha if present in the bitstream and the 2-D Discrete Cosine Transform (DCT) coefficients(, )lkX uv , prior to the IDCT.1 The variable, l, indicates the macrob

31、lock index. The variable, k, indicates the DCT block index. The variables u and v represent the horizontal and vertical indices, respectively. The ordering of the output data shall follow the raster subdivision into macroblocks (see Figure 2 to Figure 8 as appropriate in SMPTE ST 2019-1). That is, 0

32、,1,., TlN, where TN is the number of macroblocks in a frame or field. Within a macroblock, the ordering shall be according to the DCT order as per Figure 9 to Figure 11 and Figure 15 in SMPTE ST 2019-1. In other words, 0,1,.,3k , 0,1,.,5k , 0,1,.,7k or 0,1,.,11k . Within a DCT block, the ordering of

33、 the DCT coefficients shall be from top to bottom, left to right as a raster scan would display the image. That is the first coefficient is (0,0)lkX , followed by (1,0)lkX . The last coefficient within a DCT block is (7,7)lkX . The DCT coefficient data within the diagnostic output files shall be rep

34、resented by twos complement, 16-bit values (i.e. short) in little endian format. A set of VC-3 compressed bitstreams and the corresponding diagnostic output files are provided with the SMPTE VC-3 test materials. 5.1.2 Reference Channel Output When operating in the “Reference Channel Output” mode, th

35、e reference decoder will perform the entire decoding process described by SMPTE ST 2019-1. This mode shall be used to verify bitstream conformance as detailed in Section 6.2. The “Reference Channel Output” mode will verify the contents of the VC-3 header, compressed payload, and EOF area. If a non-c

36、onforming element is found in these regions, the reference decoder will output a diagnostic error message and terminate operation. 1 The 2-D DCT coefficients should not be confused with the Cosine Basis Function values that are part of the IDCT. These coefficients are the 2-D, 8x8 block DCT transfor

37、m representation of the encoded video raster. Header Parsing Entropy Decoding Inverse Quantization VLC Codeword Tables CompressQuantization Weighting Tables VC-3 RLE Compressed Alpha Input VC-3 Compressed bitstream Diagnostic output File SMPTE RP 2019-2:2016 Page 7 of 24 pages When operating in the

38、“Reference Channel Output” mode, the reference decoder will utilize a floating point implementation of the IDCT to avoid any problems associated with numerical precision such as arithmetic underflow or overflow. Conforming decoder implementations need not use a floating point IDCT implementation. Su

39、itable integer IDCT formulations exist. The output file format for the reference decoder in this mode will depend on the Compression ID. For Compression ID other than 1256 and 1270, format will be header-less binary YCBCR(A) 4:2:2(:4) or 4:2:0(:4), with components written in order as per Figure 3 an

40、d Figure 4. The samples will be presented in an interleaved order CB, Y, (A), CR, Y, (A) or planar order Plane1(Y), Plane2(CB), Plane3(Cr), (Plane4(A) . For interlaced video, full-height frames will be produced as field one followed by field two. For compressions ID 1256 and 1270, the output file fo

41、rmat will be header-less binary interleaved order CB, Y, CR, (A) 4:4:4(:4) or RGB(A) with components written in order as per Figure 5 and Figure 6. Note that alpha-components (shown in and () are optional and will be skipped (not 0 padded) if not present. The output samples of the reference decoder

42、will be arranged from left to right, top to bottom. The first sample will be the leftmost sample of the top row. This is the same order as a raster scan would display an image. Figure 3 Decoder output sequence (YCBCR(A) 4:2:2(:4) Figure 4 Decoder output sequence (YCBCR(A) 4:2:0(:4) 0 1 2 3 4 5 6 7 8

43、 9 11 12 CB A Cr A Y1 Y2 0 1 2 3 2n-1 2n Y1 Yn Y2 Plane 1: 0 1 2 3 n/2-1 n/2 CB1 CB n/4 CB2 0 1 2 3 n/2-1 n/2 CR1 CR n/4 CR2 0 1 2 3 2n-1 2n A1 An A2 Plane 2: Plane 3: (Plane 4): SMPTE RP 2019-2:2016 Page 8 of 24 pages Figure 5 Decoder output sequence (YCBCR(A) 4:4:4(:4) Figure 6 Decoder output sequ

44、ence (RGB(A) For 10-bit or 12-bit components, each will be represented by 2 bytes in ”big endian” order according to Figure 7 and Figure 8; within the 16-bit field, the most significant 10 bits or 12 bit will be valid and the other six or four lower bits will be ignored. Ignored bits will have a val

45、ue of 0. The most significant 8 bits of a component will be mapped into the first byte (byte 0) and the lower 2/4 bits will be mapped into top 2/4 bits of the second byte (byte 1). The VC-3 reference decoder will support 8-bit,10-bit and 12-bit decoding modes. The exact decoding mode will be determi

46、ned by the SBD flag (see ST 2019-1, Figure 20 for more information on the Sample Bit Depth flag). For either bit length, the output of the decoder will be one of described above formats with the lower 8 bits (byte 1) of 16 bits being zero for the 8-bit mode. Figure 7 Layout of the 10-bit data within

47、 the 16-bit data field Figure 8 Layout of the 12-bit data within the 16-bit data field 0 1 2 3 4 5 6 7 CB Cr A Y1 0 1 2 3 4 5 6 7 B R A G b9 b8 b7 b6 b5 b4 b3 b2 b1 b0 0 0 byte 0 0 0 0 0 byte 1 b11 b10 b9 b8 b7 b6 b5 b4 b3 b2 b1 b0 byte 0 0 0 0 0 byte 1 SMPTE RP 2019-2:2016 Page 9 of 24 pages 5.2 Te

48、st Bitstreams and Data Files The set of bitstreams and data files that shall be used for testing a target decoders conformance are specified in Annex A.1. The bitstreams and data files are provided with the reference decoder. The set of test bitstreams is not intended to represent all possible combi

49、nations of variable length codewords and DCT coefficients. Rather, the set of bitstreams was developed according to the following guidelines: The test bitstreams will follow the bitstream syntax and range of permissible values as defined in SMPTE ST 2019-1. Parameters specific to the display process will not be tested, as the display process is outside the scope of the SMPTE ST 2019-1 standard. There will be no tests that are frame rate dependent.