ABSTRACT
Video compression technologies play a key role in the distribution of video content in broadcasting.
Many techniques have been proposed to improve the coding efficiency provided by the most recent video compression standard, High Efficiency Video Coding (HEVC). This includes the latest tools being investigated within the Joint Video Exploration Team (JVET) and the tools provided by the royalty-free AV1 codec developed by the Alliance for Open Media.
This paper analyses the overall compression capabilities of these two emerging technologies with respect to HEVC, including objective and subjective performance evaluation and the associated encoding and decoding complexities.
INTRODUCTION
Due to the increasing consumption of video content with higher resolutions, the need for more efficient video compression techniques is growing.
The first version of the High Efficiency Video Coding (HEVC) standard (1), jointly developed by the ITU-T VCEG and the ISO MPEG, was finalised in 2013. A wide range of products and services support HEVC for video encoding/decoding, especially for Ultra High Definition (UHD) content, where HEVC can provide around 50% bitrate savings for the same subjective quality as its predecessor H.264/AVC (2).
However, HEVC is not yet as widely adopted as H.264/AVC. As an alternative to HEVC, the Alliance for Open Media (AOMedia) initiated the development of the AOMedia Video 1 (AV1) specification (3). AOMedia was founded in 2015 as a consortium of partners from the semiconductor industry, video on demand providers and web browser developers, specifically to create an open, royalty-free video coding specification. AV1 was built using Google’s VP9 specification (4) as a base, and similarly to VP9, AV1 follows the typical hybrid block-based approach also used in the standards from the MPEG/ITU family.
The AV1 specification was primarily finalised at the end of March 2018, with some further minor details defined shortly after. At the same time as the above, work on video compression technologies beyond the capabilities of HEVC continued by the MPEG/ITU, with the creation of the Joint Video Exploration Team (JVET) on future video coding in October 2015.
Many new coding tools have been proposed in the context of JVET, which eventually led to a Call for Proposals on video coding technologies with video compression capabilities beyond HEVC. The reference software used in the exploration phase of JVET, called Joint Exploration Model (JEM) (5), was leveraged as the base for the majority of responses to the call. Results included responses demonstrating compression efficiency gains of around 40% or more with respect to HEVC (6). This initiated the work by the Joint Video Experts Team (JVET) on the development of a new video coding standard, to be known as Versatile Video Coding (VVC).
Even though JEM was created as an exploration software model, it provides clear evidence of video coding technology that can significantly outperform the capabilities of HEVC. Many of its tools are therefore likely to be included in the final version of the VVC specification, expected to be completed by the end of 2020. Given these new trends, this paper provides a performance comparison between HEVC, AV1 and the JEM software, based on both objective and subjective tests.
This analysis provides indicators on how the achieved objective bitrate savings translate into subjective quality advantages. Moreover, the associated encoding and decoding complexity of both technologies is also analysed with respect to the HEVC reference software.
No comments yet