AVS2视频编码码率控制算法

发布时间：2018-05-22 08:41

本文选题：码率控制 + 模糊控制器　；参考：《系统工程与电子技术》2016年09期

【摘要】：码率控制是视频图像编码技术中的重要组成部分,旨在受限的通信带宽下提供最小失真的图像质量同时保证比特流的平稳输出,长期以来既是研究重点也是应用热点。面向中国自主知识产权的新一代音视频编码标准(audio video coding standard 2.0,AVS2)中码率控制技术,将模糊控制理论与码率控制相结合,提出了一种新的码率控制算法。首先根据模糊控制器原理建立快速查询表,接着基于指数率失真模型进行了基本论域到模糊子集论域的映射,而后将假设参考解码缓冲区反馈值及其差值带入模糊控制查询表得出量化参数(Quantization Parameter,QP)值变化增量。最终的编码过程采用系统QP值与增量之和来进行。在通用测试条件下的实验结果表明,新算法比特率失真比率(Bj錙ntegaard delta bit rate,BDBR,BD-rate)取得增益效果明显,比特率控制精度高。算法实现简洁,计算复杂度低,特别适合恒定目标比特率编码端速率控制。相关成果已被AVS2标准工作组采纳,集成在参考模型软件RD10.0中。
[Abstract]:Rate control is an important part of video coding technology. It aims at providing minimum distortion image quality under limited communication bandwidth and ensuring the steady output of bit stream. It has long been the focus of research and application hotspot. This paper presents a new rate control algorithm based on fuzzy control theory and rate control, which is a new audio / video coding standard for China's own intellectual property rights (Audio video coding standard 2.0 / AVS2). Firstly, a fast query table is built according to the principle of fuzzy controller, and then the mapping from the basic domain to the fuzzy subset domain is carried out based on the exponential rate-distortion model. Then, the feedback value and its difference of the reference decoded buffer are introduced into the fuzzy control query table to obtain the quantization parameter change increment. The final coding process is based on the sum of QP values and increments of the system. The experimental results under the general test conditions show that the new algorithm has obvious gain effect and high bit rate control precision. The algorithm is simple in implementation and low in computational complexity. It is especially suitable for constant target bit rate coding end rate control. The related results have been adopted by the AVS2 Standards working Group and integrated into the reference model software RD10.0.
【作者单位】：电子科技大学计算机科学与工程学院;电子科技大学机器人研究中心;
【基金】：国家高技术研究发展计划(863计划)(2015AA015903) 国家自然科学基金(61571102) 四川省科技支撑计划(2014GZ0005) 中央高校基本科研业务费(ZYGX2015J065)资助课题
【分类号】：TN919.81

【相似文献】