视频压缩码流在传输过程中经常会遇到因各种原因引起的误码丢包等网络传输错误。由于目前常用的编码器普遍采用“预测-残差”方式进行编码，因此一旦传输发生错误，不仅会影响当前帧的重建，更会造成空间及时间上的误差蔓延。在传输系统中，较为实际的做法是由系统各个部分共同分担这种抗网络传输差错的“责任”，通过前后部分之间不断的呼应协作，合理地满足用户对于视觉质量、实时性等各个不同方面的QoS要求。为了对抗网络传输错误对视频重建质量的影响，在结合了H．264容错编码、信道保护、打包策略、拥塞控制、差错掩盖等算法的基础上，提出了一个集成拥塞控制的端对端视频传输系统框架，并对其中关键算法进行了分析与设计，实验证明，该视频传输系统框架在不同网络情况下，均可收到较好的效果。

When transmitting over the error-prone channels, the compressed video bit-steams are always vulnerable to all kinds of transmission errors. The corrupted parts of the bit-streams may not only affect the reconstruction effect of the current frames, but also cause the error propagation problems both spatially and temporally. So it is really practical to take into account the error-resilience effects in each step of the transmission system. In this paper, we describe a framework for video transmission over the packet-loss channel which features the coordinated operation of the H. 264 error resilience encoding, error concealment algorithms, transport-layer packetization and congestion control mechanisms. We analyze and design the key algorithms of the framework and the experimental results show that our approach works well in different channel conditions.