Lagrange multiplier based perceptual optimization for high efficiency video coding
Lagrange multiplier based perceptual optimization for high efficiency video coding
复制标题
DOI：
10.1109/apsipa.2014.7041596
复制
发表时间：
2014-12
期刊：
影响因子：
--
通讯作者：
Shanshe Wang;Siwei Ma;Debin Zhao;Wen Gao
中科院分区：
文献类型：
--
作者：
Shanshe Wang;Siwei Ma;Debin Zhao;Wen Gao
本文提出了一种基于拉格朗日乘子的感知优化方案，用于高效率视频编码（HEVC）。首先，我们提出了一种非线性调整方案的拉格朗日乘子的基础上专门的编码结构的HEVC和人类视觉系统（HVS）的特点。然后将拉格朗日乘子引入到率失真优化中进行模式选择。实验结果表明，该方法能显著提高视频的主观质量。与HM锚相比，随机接入、低延迟P和低延迟B测试配置的最大主观质量改善可高达30%，平均主观性能改善分别超过7%、11%和10%。
In this paper, a Lagrange multiplier based perceptual optimization scheme is proposed for high efficiency video coding (HEVC). Firstly, we propose a non-linear adjustment scheme for the Lagrange multiplier based on the specialized coding structure of HEVC and human visual system (HVS) characteristic. Then the Lagrange multiplier is incorporated into the rate distortion optimization for mode decision. Experimental results demonstrate the proposed scheme can significantly improve the subjective video quality. Compared to the HM anchor, the maximum subjective quality improvement can be up to 30% and the average subjective performance improvement is over 7%, 11% and 10% for random access, low delay P and low delay B testing configurations respectively.