基于H.264/AVC的码率控制算法研究

【作者】 黄莉；

【导师】 傅志中；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2007， 硕士

【摘要】 2003年3月,联合视频工作组JVT(Joint Video Team)正式推出了新一代视频压缩编码标准H.264/AVC(Adavanced Video Coding),该标准是ITU-T和ISO/IEC两个国际标准化组织继MPEG-2/H.262标准后,再一次共同推出的视频编码标准。H.264/AVC标准支持低带宽、高误码率的无线移动视频通信,高码率、低延迟的视频广播,及在线流媒体等多种应用,因此日益受到业界的广泛关注。本文首先讨论H.264/AVC的系统组成和原理,及H.264/AVC的帧内预测、帧间预测、熵编码和去方块滤波等关键技术。H.264/AVC与原有一些视频编码标准相比,在系统组成上没有变化,而是通过对系统模块的改进达到编码效率的提高。其次,本文介绍码率控制算法。码率控制是视频编码的关键技术之一,其控制效果的好坏直接影响到码流的带宽和解码图像质量。由于图像序列中各帧图像的复杂度和运动性大小的非平稳性,导致其压缩编码后的输出码流产生很大的波动,所以需要采用适当的码率控制策略将这种波动控制在一个限定的范围之内。现有的视频压缩国际标准对码率控制部分是开放的,可以根据不同的情况采用不同的码率控制算法,该领域也是目前的研究热点。现今既有的经典码率控制算法包括TM5,VM和TMN8等,在本文中都有简要的介绍。H.264/AVC JM10.1码率控制算法和RDO模型密切相关,因此本文特别介绍了H.264/AVC在宏块编码模式选择时,采用的基于Lagrange优化算法的RDO(率失真优化)模型。然后本文分别从GOP层、帧层和基本单元层,详细介绍了H.264/AVC JM10.1码率控制算法。最后本文重点研究H.264/AVC JM10.1码率控制算法,深入讨论了其存在的三点不足:一、比特分配未考虑图像复杂度,二、对预分配比特和实际产生比特不匹配情况调整不足,三、在视频序列中存在场景切换帧时未作出有效应对措施,易造成缓冲溢出和跳帧。针对这三点不足之处,本文分别从基于图像复杂度的比特分配、基于RDO模型的宏块编码比特数调整和场景切换帧检测三个角度讨论了三种码率控制改进算法。原有的JM10.1编码平台没有采纳跳帧策略,本文在JM10.1中融入了跳帧代码,并在此平台上仿真验证了原算法和三种改进后算法。仿真结果表明,改进后算法与原算法相比,在平滑解码图像质量、缓冲控制和避免跳帧上都有一定的改善。更多还原

【Abstract】 H.264/AVC video coding standard is developed formally by JVT(Joint Video Team) in March 2003, and it will be the second successful video compression standard after the MPEG-2/H.262. The standard is designed to give a new technical solution to a broad range of applications, such as very low bit rate conversational services and low delay broadcast, interactive video services etc. So this standard obtains much attention from academic and industrial fields.Firstly, the system and key techniques in H.264/AVC, which including transform, inter-prediction, intra-prediction, entropy coding and block filtering, are discussed,. Compared to existing video coding standards, H.264/AVC’s improvements are mainly on the system blocks, but not the whole system.Secondly, Rate control algorithm is discussed. Rate control is one of the key techniques for video coding. It will affect the bandwidth of generated bits and the reconstruct image quality directly. Because the complexity and motility of images are not the same, it will lead to the output bits change obviously. So it is necessary that the change is controlled to a limited range by some rate control algorithms. Three classical rate control algorithms, including TM5, VM8, and TMN8, are introduced in this paper. Because the RDO mode is related to the H.264 JM10.1 rate control algorithm, The RDO(Rate Distortion Optimization)mode, which is based on Lagrange optimization algorithm and used to determine the micro block’s coding mode, is especially discussed. And then JM10.1 rate control algorithm which including GOP layer rate control, frame-layer rate control and basic unit layer rate control is elaborated.Lastly, by noticing that the JM10.1 rate control algorithm has some disadvantages such as, in the bit allocation process the picture complexity is not considered, the mismatch of actual bits generated by the coded frame and the bits allocated for this frame isn’t well adjusted and the buffer bit occupation level isn’t controlled well when the scene changing frames exist in the video sequence. Aiming at these disadvantages, three rate control improvement methods are discussed, they are seperately based on the bit allocation, the rdo mode and detection of scene changing frames. All simulations are performed on the JM10.1 platform which including the frame skip scheme. The simulation results show us, the improved algorithms do have some advantages over the JM10.1 rate control algorithm, such as the buffer controlling and the smoothness of the decoded sequence.更多还原