【作者】 韩慧；

【导师】 陈贺新；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2005， 硕士

【摘要】 可伸缩性视频编码以其良好的带宽适应能力和可提供分级保护措施等优点成为解决无线视频通信诸多难题的优良应对方案。已有的可伸缩性方法包括时域、空域、SNR 等提供固定增强层速率的经典可伸缩性方法,及其改进,提供可任意截断增强层码流的精细可伸缩性方法(FGS)。这些方法可以提供一定的可伸缩性能力,但其不足在于它们都是单域方法,它们都只提供某一个视频参数的可伸缩性,性能单一、针对性强、缺乏普遍的适应性,而目前复杂网络要求提供一种具有综合可伸缩性能力的方案,解决用户终端多样和编码效率均衡的问题。我们在对已有的两种混合可伸缩性编码方法进行详细分析后,提出了三域混合可伸缩性编码方法。它能同时提供可任意组合的时域、空域、SNR 三种可伸缩性,并将FGS 的位平面方法应到三个增强层,从而整合了增强层,输出统一码流。该方法即解决了用户终端多样性的问题,同时也兼顾了编码效率。本文详细阐述了三域混合可伸缩性编码的设计思想、码流框架、编码模式及其选择算法、编解码器设计以及改善其抗误码性能的方法。文章最后通过一系列实验验证了新方法的性能。更多还原

【Abstract】 The real-time video communication in wireless channels is one of the most important and challenging video utility in the future. To transmit video information in wireless channels should notice the characters of wireless channels such as narrow bandwidth, bandwidth fluctuation causing by multi-path fading, co-channel interference and all kinds of noise, high bit error rate and terminal variety in mobile networks. Scalable video coding method can solve some difficulties above. It can fit the change of bandwidth and terminal variety flexibly and it can provide some error resilience capability in high compression rate condition by provide different protection method in different layers. Its core idea is to separate the video stream to one base layer and one or more enhancement layer. The base layer includes the basic video information and the decoder received it can reconstruct the basic images. The enhancement layer includes more detail information and it can reconstruct the better images by adding the base layer. However, the past methods of scalability is based on some one parameter of video such as the temporal scalability basing on frequency, the spatial scalability basing on resolution and SNR scalability basing on image PSNR. They concentrate themselves on one field and be lack of referring and utilizing other methods. This makes them solving problems singly such as the SNR scalability can not fit the needs of different sizes of users and spatial scalability’s coding efficiency is low. To solve these problems, this article provides a new method named更多还原