Understanding Video Texture: a Basis for Video Compression

Angeliki Katsenou

University of Bristol

David R. Bull

University of Bristol

About

Although recent video coding standards such as HEVC, VP9, and AV1 have achieved impressive compression gains with significantly better rate-quality performance compared to their predecessors, they are all challenged by certain types of content, in particular complex dynamic textures. The VVC standard adopts a similar coding architecture and, while offering overall coding gains, it still exhibits the same limitations. A recent statistical analysis of HEVC reference software performance has shown that the codec handles various types of texture very differently in terms of coding modes and bit rate. Knowledge about the compression characteristics of video content prior to encoding can be exploited in various situations, including: off-line rate-quality optimisation of video-on-demand streamed content, multi-pass encoding, rate control, statistical multiplexing, in loop rate-distortion optimisation.

Within this project we have defined the different types of video texture, we have analysed their spatio-temporal features, we studied their coding perfromance and statistics. The collection of papers below provide all related studies. We also created a dataset that helped us with these studies, BVI-HomTex.

Funders
FP7 Marie-Curie Integrated Training Network PROVISION
Collaborators
Fraunhofer HHI, BBC, RWTH Aachen University, University of Nantes

Visuals

Examples of Video Texture

Downloads

  • Papers
    • A. V. Katsenou, M. Afonso, and D. R. Bull, "Study of Compression Statistics and Prediction of Rate-Distortion Curves for Video Texture". [arXiv]
    • A. V. Katsenou, Th. Ntasios, M. Afonso, D. Agrafiotis and D. R. Bull, "Understanding video texture — A basis for video compression," 2017 IEEE 19th International Workshop on Multimedia Signal Processing. [Paper]
    • A. V. Katsenou, M. Afonso, D. Agrafiotis and D. R. Bull, "Predicting video rate-distortion curves using textural features," 2016 Picture Coding Symposium. [Paper]
    • M. Afonso, A. V. Katsenou, F. Zhang, D. Agrafiotis and D. R. Bull, "Video texture analysis based on HEVC encoding statistics," 2016 Picture Coding Symposium. [Paper]
  • Dataset [VIL-HomTex]