基于H.264的FGS视频编码算法研究

【作者】 饶琴；

【导师】 林其伟；

【作者基本信息】 华侨大学 ， 通信与信息系统， 2011， 硕士

【摘要】 随着网络多媒体的发展,视频编码的目标从面向存储转为了面向传输,这对视频编码技术和视频网络传输技术都提出了新的挑战,其中网络带宽的波动问题是需要解决的一个基本问题。传统的用于解决该问题的视频编码方法有联播和码流转换。但这两种方法的码率只能在几个特定的码率中变化,难以实现传输过程中任意码率的动态切换,从而不能精细的匹配网络带宽的变化。为解决这个问题,精细粒度可分级视频编码技术应运而生。使用该技术编码后的视频流能很好的匹配网络带宽的变化,然而这个特性的获得是以降低编码效率为代价的。为提高FGS视频编码的效率,本文提出了三种改进方案。首先,分析了单环算法的优缺点,并在保留其将增强层信息引入到基本层,通过使用质量更高的扩展基本层图像用作基本层的参考图像,来提高基本层和增强层编码效率的优点的基础上,本文提出了两种提高感兴趣区域编码效率及改善视频主观质量的方法。利用有效的码流资源和信道资源,引入肤色选择算法和宏块分割算法来自适应的选择感兴趣区域,通过将该区域的宏块进行位平面提升、优先编码和传输,来提高感兴趣区域的编码效率及改善解码后视频的主观质量。其次,分析了双环算法的优缺点,并在保留其增加增强层的运动补偿环路来提高增强层编码效率的优点的基础上,针对其编码复杂度高和基本层编码效率低的缺点,本文提出了基于双环改进的FGS视频编码方法,该方法通过将基本层和增强层的运动估计模块进行合并,来降低编码器的编码复杂度,减少编码时间,同时将能够正确解码的增强层信息引入基本层,通过使用质量更高的扩展基本层图像用作基本层的参考图像,来提高基本层的编码效率,并进一步提高增强层的编码效率,改善解码后视频的质量。通过PC机的实验仿真结果可以看出,本文提出的改进方案效果明显,不仅提高了FGS的编码效率,也较好的改善了解码后视频的主观质量。更多还原

【Abstract】 With the development of multimedia network, the goal of video coding have been changed from storage-based to transmission-based, it put forward a great challenge for current video coding technology and network transmission technology, among them, the fluctuation of network bandwidth is a fundamental problem need to be solved. The traditional scalable video coding mathod used to solve the problem are Broadcast and Streaming-conversion, but the bitrate of these two methods are restricted in several specific bitrate, and difficult to switch it dynamic during the transmissing, so it couldn’t match the change of network bandwidth very well. To solve these problems, fine granularity scalable video coding technology arises at the historic moment. The video which coded by FGS technology can match the change of network bandwidth very well, but this will reduce the coding efficiency.To improve the efficiency of FGS video coding, three methods are put forword in this paper. First, analysed the relative merits of one-loop arithmetic, and in the base of keep the information of enhance layer add to the base layer which can increase the coding efficiency through use the higher quality patulous base layer, we put forward two methods to increase the coding efficiency of ROI(region of interested) and improve the subjective quality. Using the effective channel resources and stream resources, we proposed the methods of introducting skin-color choice algorithm and macroblock segmentation algorithm to selective the region of interest adaptive to increase the coding efficiency of ROI and improve the subjective quality by upgrading the bit-panle、preferred coding and transmission the ROI at the same time. Secondly, analysed the relative merits of two-loops arithmetic, and in the base of keep the loop of motion compensation to increase the efficiency of enhence layer, we proposed the method of improvement of FGS which based on the two-loops, in this method, we merge the module of motion estimation of the enhance layer and base layer to reduce the coding complexity, coding time and introduct the enhance layer information which can decode correctly to base layer to increase the coding efficiency and improve the subjective quality of video through use the higher quality patulous base layer. From the simulation results of the PC machine, we can know that the methods proposed in this paper can not noly increase the coding efficiency of FGS, but also can improve the subjective quality of video.更多还原