Slice-based parallel approach for HEVC encoder

Abstract

The high efficiency video coding (HEVC) is the newest video coding standard from the ITU-T Video Coding Experts Group and the ISO/IEC Moving Picture Experts Group, which significantly increases the computing demands to encode video to reach the limits on compression efficiency. Our interest is centered on applying parallel processing techniques to HEVC encoder to significantly reduce the computational time without disturbing the coding performance behavior. We propose a parallelization approach to the HEVC encoder which is well suited to multicore architectures. Our proposal uses OpenMP programming paradigm working at slice parallelization level. We encode several slices of each frame at the same time using all available processing cores. The results show that speed-ups up to 9.8 can be obtained for the All Intra mode and up to 8.7 for Low-Delay B, Low-Delay P and Random Access modes for 12 processes with a negligible loss in coding performance.

Heuristics for tile parallelism in HEVC

“…In [9] the problem of balancing slices was tackled by introducing more slices than the number of available cores. In HEVC, the authors in [10] evaluated slice parallelism using fixed slices under various encoding scenarios, while in [4] the CTU cost estimation used for adapting slice size, was based on weighting upon the depth of each CU comprising the CTU. Last, in [11] the GOP structure for LD setting was used to estimate CTU cost.…”

Frame-Based and Subpicture-Based Parallelization Approaches of the HEVC Video Encoder

“…In [14], the authors compare the encoding performance of slices and tiles in HEVC. In [15], a parallelization of the HEVC encoder at slice level is evaluated, obtaining speed-ups of up to 9.8x for All Intra coding mode and 8.7x for Low-Delay B, Low-Delay P, and Random Access modes, using 12 cores with a negligible rate/distortion (R/D) loss. In [16], two parallel versions of the HEVC encoder using slices and tiles are analyzed.…”

Algorithms and advanced methods for video coding in large scale systems

“…In [30] the problem of balancing slices was tackled by introducing more slices than the number of available cores. In HEVC, the authors in [44] evaluated slice parallelism using fixed slices under various encoding scenarios, while in [40] the CTU cost estimation used for adapting slice size, was based on weighting upon the depth of each CU comprising the CTU. Last, in [45] the GOP structure for LD setting was used to estimate CTU cost.…”

“…Then, each frame in a GOP is encoded using slice-level parallelism. In HEVC, the authors of [44] evaluated slice-based parallelism under different encoding scenarios. However, load balancing slices was not taken into account.…”

“…Efficient video coding is related to parallelism since the emergence of MPEG-2 back in the 90s, see for instance [16]. The aforementioned three parallelization options, i.e., slices, tiles and wavefront, have seen considerable research effort in the recent years, e.g., [26] (WPP), [44], [45] (slices), and [46] (tiles). All Chapter 8: On Planning the Adoption of New Video Standards in Social Media Networks: A General Framework and its Application to HEVC these parallelization approaches are rather coarse-grained, meaning that parallelism occurs at the level of groups of CTUs.…”