当前位置: X-MOL 学术Signal Image Video Process. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A multicriteria optimization of the discrete sine transform for versatile video coding standard
Signal, Image and Video Processing ( IF 2.0 ) Pub Date : 2021-06-21 , DOI: 10.1007/s11760-021-01925-2
Sonda Ben Jdidia , Fatma Belghith , Maher Jridi , Nouri Masmoudi

The additional discrete transforms from cosine (DCT) and sine (DST) families that come with the Adaptive Multiple Transform (AMT) approach is one of the major enhancements involved in the new Versatile Video Coding (VVC) standard. They have increasingly brought additional complexity compared to HEVC standard. The transform module is one of the most consuming stages in terms of time and hardware resources. This paper focuses on the optimization of the DST-VII transform. It deals with a multicriteria optimization algorithm for approximate computing that aims at the identification of the optimal approximation of the DST-VII according to several approximation measures. The resulting transform matrix has extremely low arithmetic complexity as well as close proximity to the exact DST-VII. Moreover, hardware synthesis results denote that the simplified design of DST-VII consumes only a third of the hardware resources used by the original algorithm. Experimental results obtained from joint exploration model simulations show a slight bit-rate increase while maintaining almost the same video quality. Such results confirm the effectiveness of the proposed approximate transform. The latter yields good performance in terms of computational complexity reduction and proximity to the exact transform while exhibiting a video coding performance comparable to the original algorithm.



中文翻译:

用于通用视频编码标准的离散正弦变换的多准则优化

自适应多重变换 (AMT) 方法附带的余弦 (DCT) 和正弦 (DST) 系列的附加离散变换是新的多功能视频编码 (VVC) 标准中涉及的主要增强功能之一。与 HEVC 标准相比,它们越来越多地带来了额外的复杂性。就时间和硬件资源而言,转换模块是最耗费时间的阶段之一。本文重点研究DST-VII变换的优化。它涉及一种用于近似计算的多准则优化算法,旨在根据几种近似度量确定 DST-VII 的最佳近似。生成的变换矩阵具有极低的算术复杂度,并且非常接近于精确的 DST-VII。而且,硬件综合结果表明,DST-VII 的简化设计仅消耗原始算法使用的硬件资源的三分之一。从联合探索模型模拟中获得的实验结果显示比特率略有增加,同时保持几乎相同的视频质量。这样的结果证实了所提出的近似变换的有效性。后者在降低计算复杂度和接近精确变换方面产生了良好的性能,同时表现出与原始算法相当的视频编码性能。这样的结果证实了所提出的近似变换的有效性。后者在降低计算复杂度和接近精确变换方面产生了良好的性能,同时表现出与原始算法相当的视频编码性能。这样的结果证实了所提出的近似变换的有效性。后者在降低计算复杂度和接近精确变换方面产生了良好的性能,同时表现出与原始算法相当的视频编码性能。

更新日期:2021-06-21
down
wechat
bug