Improved Video Compression Efficiency Through Flexible Unit Representation and Corresponding Extension of Coding Tools

Abstract

This paper proposes a novel video compression scheme based on a highly flexible hierarchy of unit representation which includes three block concepts: coding unit (CU), prediction unit (PU), and transform unit (TU). This separation of the block structure into three different concepts allows each to be optimized according to its role; the CU is a macroblock-like unit which supports region splitting in a manner similar to a conventional quadtree, the PU supports nonsquare motion partition shapes for motion compensation, while the TU allows the transform size to be defined independently from the PU. Several other coding tools are extended to arbitrary unit size to maintain consistency with the proposed design, e.g., transform size is extended up to 64×64 and intraprediction is designed to support an arbitrary number of angles for variable block sizes. Other novel techniques such as a new noncascading interpolation filter design allowing arbitrary motion accuracy and a leaky prediction technique using both open-loop and closed-loop predictors are also introduced. The video codec described in this paper was a candidate in the competitive phase of the high-efficiency video coding (HEVC) standardization work. Compared to H.264/AVC, it demonstrated bitrate reductions of around 40% based on objective measures and around 60% based on subjective testing with 1080p sequences. It has been partially adopted into the first standardization model of the collaborative phase of the HEVC effort.

Implementation of fast HEVC encoder based on SIMD and data-level parallelism

“…Compared to cases without parallelization, four-slice parallelization for an entire frame with HM 9.0 encoder yields an ATS gain of 70.06% with only 1.52% BD-BR increase and 0.037 dB BD-PSNR reduction for class B and an ATS gain of 69.21% with only 3.36% BD-BR increase and 0.128 dB BD-PSNR reduction for class C, respectively. Figure 9 illustrates average speed-up factors of the slice parallelization, in terms of the number of slices (2,4,8,16, and 32). As shown in Figure 9, the speed-up also increases up to 6, as the number of slices increases.…”

“…A CTU is split into multiple coding units (CU), in a quadtree fashion [3]. Along with the CU, the prediction unit (PU) and transform unit (TU) are defined, and their sizes and shapes are more diverse than the prior standard technologies [4,5]. On top of them, many advanced coding tools that improve prediction, transform, and loop filtering are employed to double the compression performance compared with H.264/AVC.…”

An Simplified Reference Frame Selection Configuration Method

“…The reference frames of RPS are classified into two groups: shortterm reference frame set, and long-term reference frame set [11]. Short-term reference frame set can be classified into five subsets.…”

Discrete Sine Transform-Based Interpolation Filter for Video Compression

“…Therefore, the finite filter lengths of interpolation filters are determined and motion vectors are supported with 1/4-pixel accuracy in HEVC. During the development of HEVC, there were several proposed interpolation filter techniques, such as switched interpolation filters with offset (SIFOs) [11], maximum order of interpolation with minimal support (MOMS) [12], one-dimensional directional interpolation filters (DIFs) [13], and DCT-based interpolation filters (DCT-IFs) [14]. As a result, the DCT-IFs are adopted in HEVC for the sake of coding efficiency.…”