【作者】 张梦晗；

【导师】 殷福亮； 陈喆；

【作者基本信息】 大连理工大学 ， 信号与信息处理， 2009， 硕士

【摘要】 随着无线网络技术和音视频技术的发展,移动多媒体编解码应用在人们的生活中扮演越来越重要的角色。为了支持不同编码标准之间的设备能够相互通信,需要对已经编码的码流数据进行相互的转换,这一转换过程就叫做视频转码。AVS-M是我国专门针对移动多媒体应用自行制定、拥有自主知识产权的数字音视频编解码标准,为AVS系列编解码标准的第七部分;H.264-Baseline为H.264/AVC(Advanced Video Coding,先进视频编码)标准的基本档次,为国际移动多媒体应用的主流标准。本文的研究工作围绕两种标准相互间的转码展开。本文的主要工作如下:(1)对H.264-Baseline和AVS-M两种视频编解码标准的基本技术要点进行了简要介绍,并通过对比两者之间的不同,提出了实现转码器所需解决的主要问题。(2)通过对H.264-Baseline和AVS-M之间整数变换方面的比较,本文提出了一种基于变换域的变换系数转码快速算法,该算法通过简单的蝶形计算将两种码流的变换系数矩阵互转,结合量化过程实现残差数据的快速转码。(3)通过对帧内预测编码部分的比较分析,本文提出了一种帧内预测模式的快速映射算法,该算法结合16×16亮度宏块切割模式映射等方法,重用了码流中包含的帧内模式预测结果,极大程度的提高了帧内最佳预测模式的判断速度。(4)通过对帧间预测编码部分的比较分析,本文提出了一种树形结构的帧间预测转码算法,该算法针对不同的宏块选择性进行运动矢量预测的修正和插值数据残差的重计算,在保证图像质量的前提下提高帧间转码的效率。(5)将本文提出的算法应用于转码器的构建程中,利用一系列测试完成对转码器的转码质量和效率测试。结果表明,该转码器能够正确实现H.264-Baseline和AVS-M之间的转码,并能在保证图像质量的同时大大提高转码效率。更多还原

【Abstract】 With the development of wireless network and audio and video technology, the application of mobile multimedia codec plays an important role more and more in social life. For supporting the communication between different sets with different code standard, it is necessary for transforming the coded bit-stream into another style. This process is called video transcode. As a digital audio and video codec standard established and owned by China, AVS, Audio and Video Standard, includes a video standard called AVS-M as its seventh part; H.264-Baseline, the base level of H.264/AVC(Advanced Video Coding), is the mainstream standard of international mobile multimedia communication. The essay’s work is the transformation between them:(1) Introducing the principle of H.264-Baseline and AVS-M, comparing with each other and taking the keys about realizing the transcoder.(2) After comparing the part of integer transform between H.264-Baseline and AVS-M, the essay presents a transcoding fast algorithm about transform coefficient in the transform field, which transforming the transform coefficient matrix between them by using butterfly-type algorithms, realizes the remnant data ’s fast transcoding.(3) After analyzing and finding difference between the part of intra predication, the essay presents a fast algorithm for mapping the intra prediction modes. Using the way of cutting the 16×16 luminance macro block and mapping its prediction mode, it is nice for reusing the results about intra mode prediction, advancing the speed of getting the best prediction mode.(4) After comparing the part of inter predication, the essay proposes a tree-type inter prediction transcoding algorithm, which using the motion vector correction and the regain of insert-value, advances the efficiency of inter prediction transcoding and keeps the quality of image.(5) At last, the essay builds a transcoder by using such three algorithms, tests the quality and efficiency of transcoding. This architecture can realize the transcoding between H.264-Baseline and AVS-M, and advance the efficiency of transcoding that kept the quality of image.更多还原