Research On Robust Image/Video Transmission

With the emergence of the third-generation (3G) mobile communication system and quick development of IP network, image/video communication is becoming one of major communication services. However, wireless mobile channel and IP network are highly prone to error. Channel bandwidth is valuable and limited resource, compression techniques are generally used to reduce transmission data volume. These compressed data are very sensitive to error. Transmission errors not only corrupt the current decoded data, but they may also propagate to succeeding data. It is necessary to find effective method(s) to solve the problem. At current stage, it is still an active and challenge research topic. In order to ensure quality image/video at the receiver port, multiple error resilient techniques need to be used to eliminate or reduce error. The main focus of this dissertation is to develop and test whether use multiple error resilient techniques are better than use single method on eliminating/reducing errors. The main work is as follows:1. An error resilient algorithm to collaborate multiple description coding (MDC) and error concealment (ECN) is presented. The strategy of this method is to reconstruct the low quality image followed by recovering the lost wavelet tree with error concealment method when end-user receives few descriptions. Our experiment results indicate that this method effectively improve robustness of image transmission. In addition, every description is embedded bit-streams, which is also feasible to quality scalability.2. In this paper, a gray video MD-3D-SPIHT algorithm is proposed, which extends the MD-SPIHT algorithm into 3D-SPIHT. This method is to divide the 3-D wavelet coefficients into some number P of different groups according to their spatial and temporal relationships, and then to encode each group independently using 3D-SPIHT algorithm, so that P independent embedded 3D-SPIHT substreams are created. Multiple descriptions are created. Further, combat channel errors algorithm of collaborating MDC and ECN is presented for video. In our simulation tests, this less sensitive source coder substantially increases channel error robustness over a wide range of BERs. In addition, this coder has the function of spatial and temporal scalability.3. Extending the previous algorithms experience to color video we create a new algorithm of joint MDC and ECN techniques. To improve coding efficiency and...