基于MV空间相关性的图像错误掩盖方法的研究

发布时间：2018-06-24 19:31

本文选题：图像错误掩盖 + 运动向量　；参考：《电子科技大学》2014年硕士论文

【摘要】：在多媒体视频技术迅速发展的现代社会,人们对于视频通信的质量要求越来越高,传统的视频压缩技术已经不能满足大众的需求。为了提供更好的视频通信质量以及更大的压缩率,H.264/AVC编解码协议应运而生。相比之前的编解码协议,H.264/AVC可在保持编码质量的前提下减少50%的编码比特率,这大大减轻了视频编码信号在网络信道上的传输压力。然而,在实际通信中,传输信道的不稳定性往往造成信号在传输过程中的丢失,从而导致终端接收到信号的失真。对于图像信号而言会造成图像的局部黑块、马赛克甚至整体图像的丢失。所谓错误掩盖,是利用解码段接收到的准确信息与丢失信息的相关关系进行内容的估计,从而达到恢复图像质量的效果。这一技术既提高了接收端图像质量也不会给传输线路带来额外的带宽压力,因此受到了相关图像传输系统及协议的青睐。本文在对相关图像错误掩盖算法进行学习和研究的基础上,针对H.264/AVC图像编码协议提出了两种快速而有效的基于MV空间相关性的图像错误掩盖方法。本文主要完成了以下几部分的工作:1.首先文章对多媒体视频编码原理及编解码标准的发展历史进行了简单的阐述,并着重从编解码的流程、采用的新技术特点以及其在现阶段的应用等对目前主流的视频编解码标准H.264/AVC进行详细的介绍。2.针对H.264/AVC现阶段使用的一些差错控制技术进行了研究,对几种主要的差错控制技术进行介绍和分析,包括编码端的差错控制技术以及解码端错误掩盖技术。本文将要着重描述的便是时域错误掩盖算法。3.研究了近年来针对H.264/AVC的几类时域错误掩盖算法,按照算法的运算特性分为几类分别进行阐述,说明其各自的优缺点并与错误掩盖技术进行了对比。4.提出了基于MV空间相关性的时域错误掩盖算法(MVSC)。该方法利用时域信息以及运动向量的空间相关性对已丢失的宏块进行恢复。实验表明,该算法能有效地对已丢失信息进行恢复,视频质量得到很大改善,有一定的实用价值。5.对提出的MVSC算法进行了进一步的改进,将运动趋势判断因子和部分恢复方法进行重新设计,提出了优化的基于MV空间相关性的时域错误掩盖算法(OMVSC),并进行了仿真实验得出实验效果,结果有进一步的提升。
[Abstract]:In the modern society with the rapid development of multimedia video technology, the quality of video communication is becoming more and more demanding, and the traditional video compression technology can not meet the needs of the public. In order to provide better video communication quality and greater compression ratio H.264% AVC codec protocol emerged. Compared with the previous coding and decoding protocol H.264 / AVC can reduce the coding bit rate by 50% while maintaining the coding quality which greatly reduces the transmission pressure of the video coding signal over the network channel. However, in practical communication, the instability of the transmission channel often results in the loss of the signal in the transmission process, which leads to the distortion of the signal received by the terminal. For image signals, local black blocks, mosaics and even whole images are lost. The so-called error concealment is to estimate the content by using the relation between the accurate information received by the decoded segment and the lost information so as to achieve the effect of restoring the image quality. This technique not only improves the image quality of the receiver but also does not bring additional bandwidth pressure to the transmission line, so it is favored by the related image transmission systems and protocols. On the basis of the study and research on the error concealment algorithm of correlated images, this paper proposes two fast and effective image error concealment methods based on MV spatial correlation for H.264 / AVC image coding protocol. This paper mainly completes the following parts of the work: 1. First of all, this paper briefly describes the principle of multimedia video coding and the history of the development of coding and decoding standards, and focuses on the process of encoding and decoding. The features of the new technology and its application at the present stage are introduced in detail to the current mainstream video coding and decoding standard H.264 / AVC. In this paper, some error control techniques used by H.264 / AVC are studied, and several main error control techniques are introduced and analyzed, including coding error control and decoding error concealment. This paper will focus on the time domain error concealment algorithm. 3. In this paper, several kinds of time-domain error concealment algorithms for H.264 / AVC in recent years are studied. The algorithms are divided into several categories according to their operational characteristics, and their advantages and disadvantages are explained and compared with error concealment techniques. A time domain error concealment algorithm (MVSC) based on MV spatial correlation is proposed. The time domain information and the spatial correlation of motion vectors are used to recover the lost macroblock. Experiments show that the algorithm can effectively recover the lost information, and the video quality is greatly improved, which has some practical value. The proposed MVSC algorithm is further improved, and the motion trend judgment factor and partial recovery method are redesigned. An optimized time-domain error concealment algorithm based on MV spatial correlation (OMVSC) is proposed.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN919.81

【参考文献】