 |
|||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||
        |
Global Standardization Activities Vol. 17, No. 6, pp. 49–52, June 2019. https://doi.org/10.53829/ntr201906gls Versatile Video Coding: a Next-generation Video Coding StandardAbstractThe scope of Subcommittee (SC) 29 of ISO/IEC (International Organization for Standardization/International Electrotechnical Commission) Joint Technical Committee 1 is coding of audio, picture, multimedia, and hypermedia information. Working Group (WG) 11, part of SC 29, is standardizing video coding, media transmission, streaming, audio coding, image/video retrieval, and genomic information coding. In April 2018, WG 11 initiated standardization of next-generation video coding called Versatile Video Coding (VVC), which is aimed at achieving higher compression, in conjunction with ITU-T (International Telecommunication Union - Telecommunication Standardization Sector) Study Group 19. This article introduces the background, target, and recent development status of VVC. Keywords: ISO/IEC JTC 1/SC 29/WG 11, MPEG, Versatile Video Coding
1. IntroductionSubcommittee (SC) 29 (Coding of audio, picture, multimedia, and hypermedia information) is a subcommittee of the International Organization for Standardization/International Electrotechnical Commission (ISO/IEC), an international standardization body, Joint Technical Committee (JTC) 1, and it has been the hub of standardization activities related to the coding of multimedia information technology [1, 2]. There are two Working Groups (WGs), WG 1 and WG 11, under the auspices of SC 29. WG 11 is known as the Moving Picture Experts Group (MPEG), and it develops the standards for technologies such as video and audio coding, multiplexing/synchronizing audio and visual streams, high efficiency coding, three-dimensional video coding, and image/video retrieval. These standards are known as MPEG series standards. The amount of Internet protocol (IP) traffic is increasing rapidly worldwide. It was reported that in 2017, the annual run rate for global IP traffic was 122 EB (exa (1018) bytes) per month [3]. IP traffic is predicted to grow at a compound annual growth rate of 26% from 2017 to 2022, and to increase threefold (387 EB per month) by 2022. IP video traffic accounted for 75% of all IP traffic in 2017, which is forecast to be 82% by 2022. Such video traffic content is already compressed (mainly by MPEG coding standards) down to a few hundredths of the original size. Thus, it is clear that not only video-based services but also network services as a whole would collapse without video coding standards. Also, with the trend for rapid growth of video traffic, more and more powerful compression techniques are necessary. 2. Latest video coding standardDevelopment of the most recent international video coding standard to date, H.265/MPEG-H HEVC (High Efficiency Video Coding), began in 2010 by the Joint Collaborative Team on Video Coding (JCT-VC) consisting of WG 11 and the International Telecommunication Union - Telecommunication Standardization Sector Study Group 16 (ITU-T SG16) Working Party 3, Question 6 (VCEG: Video Coding Experts Group). The initial version of the Final Draft International Standard (FDIS) was completed in January 2013. It is used in 4K/8K broadcasting and video transmission. More than 2 billion HEVC-compliant devices have been shipped worldwide. Since the initial version of HEVC, various extensions and amendments have been published [4]. HEVC effectively achieves compression performance twice that or more than any of the previous standards. However, due to the strong demand for higher compression, in 2014 MPEG began working on achieving an even better video coding standard, aiming at a next-generation video coding standard named Future Video Coding (FVC) at that time. At a meeting in Strasbourg, France, in October 2014, a brainstorming session was held inviting engineers from the communications, network service, and hardware fields as speakers. At that meeting, the requirements for FVC were discussed and defined as broadcasting, IP television, contribution quality video transmission, digital cinema, surveillance, mobile viewing, and virtual reality (VR), among others. In February 2015, JCT-VC began developing reference software for next-generation video coding called Key Technical Area (KTA). At a meeting in Geneva, Switzerland, in October 2015, the Joint Video Exploration Team (JVET) between ISO/IEC and ITU-T was created, and this team subsequently inherited KTA and used it to develop reference software called Joint Exploration Model (JEM). MPEG issued the requirements for the functionalities and performance of FVC in June 2016 [5]. In January 2017, JEM was confirmed to have achieved a significant performance gain compared to the HEVC test model (HM) [6]. In April 2017, a Call for Evidence was issued [7]. In October 2017, a Call for Proposal (CfP) was issued [8]. At the same time, the team was renamed the Joint Video Experts Team (keeping the same abbreviation JVET). After the meeting, 26 CfP responses from 21 organizations were submitted. All the proposals were carefully compared via a subjective viewing test, and the results were compiled and reported at a meeting in San Diego, USA, in April 2018. The new standard is named ISO/IEC 23090 MPEG-I Part 3 Versatile Video Coding (VVC). The “I” in MPEG-I stands for Immersive Media. Based on the subjective results and proposed technological details, a technical description document (Working Draft 1: WD1) and reference software (VVC Test Model 1: VTM1) were created as a starting point. 3. Current standardization status of VVC and future development planVVC will support three types of video: Standard Dynamic Range (SDR), High Dynamic Range (HDR), and 360° (VR-oriented, omnidirectional view). Various technologies are being proposed and evaluated in JVET’s exploration process. The target compression performance is a 30–50% bit-rate reduction compared to H.265/HEVC at the same subjective video quality. JVET holds standardization meetings four times a year. At each meeting, proposed coding tools are intensively reviewed from subjective/objective aspects. Decisions are made on which tools will be adopted, and the WD is updated based on the consensus of attendees with various backgrounds. At the same time, implementation of adopted tools, encoding optimization (outside the scope of standardization) and performance enhancement, and bug fixes are carried out on the reference software VTM using an open/public repository. Numerous contribution documents have been submitted so far, with 118 in April 2018 (beginning of standardization), 559 in July 2018, 690 in October 2018, 897 in January 2019, and 858 in March 2019. This is far more than the total submitted at the first five meetings on HEVC standardization. After the March 2019 meeting, meetings will be held in Gothenburg, Sweden, (September 2019), Geneva (October 2019), Brussels, Belgium, (January 2020), Alpbach, Austria, (April 2020), and Geneva (June–July 2020), as well as a meeting in Rennes, France, (October 2020) to reach the FDIS stage. The most recent reference software for VVC, VTM5.0, achieves a 33.14% bit-rate reduction (luminance (Y) Bjøntegaard Delta rate (BD-rate)), an encoding runtime of 6.71 times, and a decoding runtime of 1.03 times compared to HEVC reference software HM16.19, under the random access coding structure. More details are given in Table 1. Faster coding tools with more coding gain are demanded.
Some of the principal tools adopted in VVC so far are listed in Table 2. Among them, CST (chroma separate tree), CCLM (cross-component linear model), ALF (adaptive loop filter), AFF (affine motion compensation), MTS (multiple transform set), and DQ (dependent quantization) contribute to improving coding performance more than others. Many proposals, including those on neural network technologies, are intensively evaluated and selected at each meeting.
4. Future exploration of WG 11This article has overviewed the background of VVC standardization, the latest trends, and future plans. Other than VVC, current MPEG standardization efforts continue on video retrieval standard ISO/IEC 15938 MPEG-7 Part 15: Compact Descriptors for Video Analysis (CDVA), neural network coding, and genomic information coding standard ISO/IEC 23092 MPEG-G. As for point cloud compression (PCC) standards, Video-based PCC (V-PCC) of MPEG-I Part 5 is scheduled to reach the FDIS stage in January 2020, and Geometry-based PCC (G-PCC) of MPEG-I Part 9 is scheduled to reach the FDIS stage in April 2020. Additionally, work is underway on high-density light field video coding, 6DoF (degree of freedom) video coding with which the user walks a few steps away from a central position in a scene, and 3DoF+ video coding, with which the user does not walk in the scene (e.g., sitting on a chair) [10]. For 3DoF+, a CfP was issued at the meeting in Marrakech, Morocco, in January 2019 [11], and it is scheduled to reach the FDIS stage of MPEG-I Part 7 in July 2020. Moreover, standardization of Low Complexity Video Coding Enhancements, which enables two-layered spatial scalability [12], and Essential Video Coding (ISO/IEC 23094 MPEG-5 Part 1) [13] are under development. MPEG’s coverage and activity is ever increasing. References
|
||||||||||||||||||||||||||||
