Translated Abstract
In recent years, with the rapid development and wide application of multimedia and computer network technology, Internet applications have transformed our lives through the way we communicate and video transmission plays more and more important role in the field of network communication. Quality properties such as continuity and low delay are required in real-time transmission of video and other similar media, so that they are sensitive to loss of packets and the network bandwidth jitter. But the majority of networks are still operating in the lost-package environment and have no the guarantee of Qos (Quality of Service). Therefore, how to provide robust, reliable, highly effective as well as elastic real-time video streams in the unreliable network, is a crucial problem, and faced with numerous challenging as well.Aimed at the international latest video coding standard H.264/SVC, the paper studies the joint source and channel coding (JSCC) method. During the sending process of video, considering time delay in the actual application and computational complexity, we regard the coded scalable video code stream as a set with different importance. Then, the appropriate code stream will be selected adaptively at the transmission according to different the network status. On the foundation of traditional rectangle UEP (unequal error protection) strategy, a novel trapezoidal UEP strategy is proposed to protect the video transmitted in lost-packets network. This strategy fully considers the factors of different importance of temporal layer and quality layer and data quantity differences in the scalable code streams. Compared with traditional rectangle UEP strategy, it significantly decreases the useless redundancy added at the transmission and reduces the actual network average packet loss rate, and then it promotes the integral performance of system. Meanwhile the paper proposes a layer-aware distortion model to estimate accurately the reconstruction video quality distortion caused by the loss of quality enhancement layer and drift propagation at decoding side and allocates optimally the source-channel code-rate according the estimated distortion value. Then, the optimization problem of source and channel code rate allocation can be modeled as a multiple-choice knapsack problem with additional constraints. And the proposed heuristic algorithm can get the solution rapidly and simply. Finally, the video robust transmission system is designed and implenmented according to the above research. The experimental results and analysis show that compared with the UEP scheme which only considers channel code-rate allocation, the proposed JSCC strategy has a much better robustness and adaptability in different network conditions and can get a higher average PSNR. And also, the coding redundancy rate reduces greatly.
Corresponding authors email
