BBC R&D

Posted by Matteo Naccari, Marta Mrak on

The most efficient video compression solution for professional applications, such as production or contribution, was published last year by ISO/IEC and ITU-R as the Range Extensions (RExt) under Version 2 of the High Efficiency Video Coding (HEVC) standard. The video coding research team at BBC R&D has been actively involved in the development of H.265/HEVC, including Version 1 that covers distribution profiles. A detailed overview of the RExt for H.265/HEVC is provided in a paper co-authored by BBC R&D and recently published in the IEEE Transactions on Circuits and Systems for Video Technology (TCSVT).

The High Efficiency Video Coding (H.265/HEVC) standard is the state-of-the-art in video compression and has been developed by ITU-T and ISO/IEC in a joint partnership called the Joint Collaborative Team on Video Coding (JCT-VC). H.265/HEVC provides superior coding efficiency when compared to its predecessors and this outstanding result becomes more evident over Ultra High Definition (UHD) content as demonstrated in a recently published study. After Version 1 of H.265/HEVC was finalised in January 2013, the focus for JCT-VC moved to the definition of extensions such as Scalable, Multiview and Range Extensions (RExt). This blog introduces some of the main features of RExt, further details of which can be found in our IEEE paper.

RExt targets a broader set of video coding applications than Version 1 of H.265/HEVC and provides support and coding tools for compression of content in monochrome, 4:2:2 and 4:4:4 chroma sampling format as well as content with bit depth beyond 10 bits per pixel. Moreover, the compression efficiency of H.265/HEVC RExt has been improved for lossless coding, content in the RGB colour space, and the so-called “screen content” which refers to video with a mix of camera captured and synthetic content (e.g. text). The design principle followed throughout the whole H.265/HEVC RExt development was to include additional thoroughly evaluated coding tools that bring coding efficiency benefits. In this regard, particular attention has been devoted to any divergence from Version 1 syntax associated with each novel tool.

To support the aforementioned 4:2:2 and 4:4:4 chroma formats and high bit depth content the intra and inter coding process have been modified accordingly. More precisely:

  • Extending the interpretation of the Transform Block (TB) partitioning for 4:2:2 and 4:4:4 chroma sampling
  • Extending the derivation of quantisation matrices for chroma blocks in 4:4:4 chroma sampling
  • Modifying the derivation of the prediction mode for chroma components in 4:2:2 chroma sampling.

 To improve the compression efficiency for lossless coding, content in RGB colour space and screen content, three novel coding tools were introduced:

  • Cross-Component Prediction (CCP), where the dependency between the three colour component is exploited so that coding residuals of one component can be predicted by the residuals of the previous component in the same transform block
  • Adaptive Chroma Quantisation Parameter (ACQP), where a different quantisation can be used for the chroma component thus improving the perceived video quality
  • Residual Differential Pulse Code Modulation (RDPCM), where the residuals in a transform block can be predicted in a sample-based fashion along either the horizontal and vertical direction.

 

Moreover, existing coding tools specified in Version 1 syntax were extended to either improve the compression efficiency for screen content coding or maximise the throughput for content with high bit depth.

Often professional video coding applications require the transmission of alpha channels used during editing and post-production. Version 2 of H.265/HEVC allows the transport of these alpha planes, which are encoded as monochrome images and embedded in the bitstream as auxiliary pictures.

Different profiles of H.265/HEVC RExt rely on different tools from H.265/HEVC Version 1. Each profile requires a compliant decoder to support only a subset of the tools available in the standard so that each manufacturer can build codecs devoted to their particular core business. H.265/HEVC RExt defines 21 profiles in addition to the 3 profiles defined in Version 1. These new profiles are designed according to a layered structure so that a decoder conforming a given profile with particular bit depth and chroma subsampling is also able to decode bitstreams associated with lower bit depth and chroma formats (i.e. lower spatial resolution for chroma components). H.265/HEVC RExt profiles span a wide range of video coding applications such as magnetic resonance imaging (monochrome profiles), broadcasting contribution where the 4:2:2 chroma format is commonly used (4:2:2 profiles) and professional cameras where the content is captured at very high bitrates and bit depth (High Throughput 4:4:4 16 Intra).

When compared to the Fidelity Range Extension (FRExt) of the H.264 Advanced Video Coding (AVC) standard, H.265/HEVC RExt provides up to 36% bitrate reduction for the same video quality measured in terms of the Peak-Signal-to-Noise-Ratio (PSNR). More details about the technology and performance of H.265/HEVC professional profiles can be found in this IEEE paper co-authored by BBC R&D.

 

Topics