基于视频内容自适应拉格朗日参数选择的HEVC率失真编码优化

Rate-distortion optimization based on video content adaptive Lagrangian parameter selection in HEVC

传统视频混合编码框架采用率失真优化技术实现最优的率失真压缩性能。本文针对新一代高效视频编码标准HEVC,提出了一种新的视频内容相关的拉格朗日参数自适应优化算法。该算法研究视频时空预测编码的内容复杂度,将时间运动差异复杂度和空间梯度强度引入率失真模型,得到了视频内容复杂度的自适应拉格朗日参数。实验结果表明,与传统的量化参数相关率失真优化方法相比,该算法在没有增加额外计算复杂度的情况下,低延时-P和低延时-B配置环境下分别平均节省了3.6%和3.5%的码率,最高节省7.2%的码率,明显提高了率失真编码性能。

Traditional video hybrid coding framework uses rate-distortion optimization technology to achieve optimal rate-distortion compression performance.In this paper,a new Lagrangian parameter adaptive optimization algorithm related to video content is proposed for High Efficiency Video Coding (HEVC),a new generation of high-efficiency video coding standard.The algorithm studies the content complexity of video spatiotemporal prediction coding,and introduces time motion difference complexity and spatial gradient intensity into rate distortion model,and obtains adaptive Lagrangian parameters of video content complexity.The experimental results showed that compared with the traditional quantization parameter correlation rate distortion optimization method,the proposed algorithm saves bit by 3.6% and 3.5% on average for low delay P and low delay B encoding structures,respectively.Maximal bit rate saving can reach 7.2%.The proposed strategy can significantly improve the rate distortion coding performance without adding additional computational complexity.