Next-Generation Video Coding and Streaming

CHAPTER 6CODING PERFORMANCE OF H.262, H.264, AND H.265

This chapter provides experimental results on the coding performance of three generations of video coding standards that were produced by the MPEG/VCEG working groups. By reviewing the file-based and frame-based coding performance, we provide an in-depth comparison of the H.265/HEVC, H.264/AVC, and H.262/MPEG-2 standards. We believe that this is the first time the frame-based video coding performance has been reported. This offers useful insights into the coding efficiencies of the individual frames, which are either intracoded or intercoded, and how they respond to different types of video content. For example, the use of unidirectionally predicted frames contributes significantly to the high H.265 coding efficiency for complex video content whereas bidirectionally predicted frames are far less effective for such content. On the other hand, P frames may be coded more efficiently than B frames by H.262 for certain types of high-resolution content. The coding efficiency of the individual frame types allows the frequency of the different frame types to be optimized in a group of pictures (GOP). We also provide extensive results detailing the impact of new H.265 features on video coding efficiency. Specifically, we studied the impact of changing the size of the coding and transform units on the coding efficiency and encoding time. The factors affecting the encoder complexity and resilience are also investigated. Overall, the ...