【作者】 张翠萍；

【导师】 赵岩；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2008， 硕士

【摘要】 随着技术的发展,多媒体通信正逐步成为通信网络的主要业务。多媒体通信,如会议电视、可视电话、图像和视频检索等,其中主要的是图像和视频媒体。视频通信的数据往往是压缩编码以后的数据。当编码视频流在Internet上传输时,很容易受到信道错误的影响而丢失数据,并且误码在接收端解码时会引起扩散,使重建图像的质量在不同程度上受到该误码的影响。这主要是由于压缩编码去掉了视频图像内大量的空间和时间相关性,从而造成压缩后的图像对信道误码十分敏感,即易受到信道误码的影响。为了减少由于信道误码引起的图像失真,人们正进行着种种努力。其中,基于解码的或者和编码相结合的多种误码掩盖方法近年来引起了人们的注意。这种方法是对已发生错误的一种弥补,因而是一类比较独立的、对改进误码图像的观赏质量有很大作用的、简单易行的方法。本文提出了一种基于数据嵌入的误码掩盖方法。该方法在新一代视频编码标准—H.264上结合编码端和解码端对受损图像进行数据恢复,即在编码端中提取出一些比较重要的数据,对其进行嵌入传输。在解码端处将其提取出来,根据嵌入的信息,对搜索到的丢失宏块进行弥补,恢复图像的视觉效果。该方法与H.264的参考软件所应用的时域误码掩盖方法进行性能比较,实验结果表明,使用该嵌入方法,通过对PSNR的比较,达到了预期目的。更多还原

【Abstract】 As the emerging of 3-G mobile communication system and booming of IP-based network, video communication wins great importance and prosperous future in multimedia services network today. Some successful applications have been very popular, such as multimedia email, interactive multimedia and e-commerce etc. At the same time, the error will come into being in the wireless channel or IP channel. Because of the limit in bandwidth of channel, data of video communication is always the one which has be compressed and coded, based on that, the data is so tender that the quality of picture will be worsen in error environment. In one word, reduction of raw video data rates is one of the key problems in wireless and grouping video communication.For effective video communication, there are two ways to avoid or recover the transmitting error. One is use of the error control and data recover in traditional data communication, such as FEC, ARQ and so on; its intention is to transmit data with no error. The other is that errors recover, such as signal reconstruction and error concealment, whose aim is to recover corrupted picture vividly.Because of the characters of human eyes, people can abide some distortion of picture or video. Based on that, corrupted picture can be recovered appropriately when video bit stream have the error, Toward the recover, it is not recovered accurately, but only conceals the picture, which will look like a whole picture. So it has the applicability in practice. For the reasons, error concealment has become to the main method in the domain of video communication. For decreasing the picture distortion owing to channel error, people attempt the whole bag of tricks. Nowadays, the technique based on encode, decode or combining of encode and decode has attracted people’s attention.This paper describes the necessity and the challenges of error control for video communications, and provides an overview of various approaches that have been developed for in the current international video coding standard H.264. We emphasize the advanced coding features presented in H.264, such as multi-shape motion compensation block size, multiple reference frame prediction, fractional pixel motion compensation accuracy and context based entropy coding methods.To cope with the problem of packet loss probability and visual degradation, H.264 introduces slice-structured coding, which provides spatially-distinct resynchronization points within the video data for a single frame. A more advanced and generalized concept is provided by Flexible macroblocks ordering, which has been introduced to have the possibility to transmit macroblocks in non-scan order. This flexibility allows the definition of different patterns including slice interleaving without interrupting the inter macroblocks prediction for motion vector prediction and entropy concealment. A third error resilience concept included in H.264 is data partitioning, which can also reduce visual artifacts resulting from packet losses, especially if prioritization or unequal error protection is provided by the network. All these error resilience features in H.264 video coding standard can greatly provide more robust error resilient capabilities and keep compliance with original standard syntax.After a general analysis of the current situation of error resilience techniques in video coding, more words focus on the error concealments in the decoder. Generally speaking, the paper describes two aspects, one is the conventional error concealment, including spatial-domain, temporal-domain, frequency-domain; and the other is the up to date one. Nowadays, a lot of new technology turns up, this can be used in the video domain excellently, such as watermark. Based on that, people are thinking of error concealment making use of encoder and decoder.Based on the research above, this paper proposes an algorithm to enhance the robustness of constructed video in error prone channel. The algorithm is the error concealment based on data embedding, which, based on the new video standard-H.264, by the methods, the decrease between the same macroblocks of frames near by will be embedded in transform coefficients. Then in the decoder, when the program has searched the lost macroblocks, the embedded data is extracted from video, we use the information to find out the best block for concealing corrupted picture.The paper suggests two embedding methods, the one is embedded by the odd-even embedding; the other is embedded by fixed points embedding. The result shows that, compared with the algorithm in test model JM72, the two methods have the more cost in transmitting channel, and the PSNR also reduces in encoder. But compared with odd-even method with the least 0.10 dB, the other makes the less decrease: the least is only 0.01 dB, for the reasons, we ignore the PSNR reducing. Towards the PSNR in the decoder, two methods have the same results more or less, but the second method makes the less embedding cost. In one word, the method of embedded by fixed points embedding has the better effect.In conclusion, the proposed temporal error concealment algorithm has the effective improvement on error concealment. Besides, the performance is even superior using proposed algorithm as the QP or acuity-motion is increasing.更多还原