Research On Wavelet Video Coding And Dynamic Bit-rate Transmission

This dissertation focuses on the video transmission applications under the uncontrollable network. It aims to provide the best video experiences to users in the network environment that the band-width changes frequently. After the contrast analysis on each kind of wavelet video coding algorithm, the most suitable wavelet coding framework is selected and the optimization research is carried based on this framework. The dynamic bit-rate transmission control algorithm is brought forward after evaluating in each process such as coding, transmission, decoding etc. It can enhance the adaptation ability of the video applications when the network environment changes and improve the video experience of users.An improved multi-frame prediction and no update barbell lifting motion compensated temporal filtering algorithm is brought forward and a fast motion estimating method of the multi-frame prediction based on the block motion model is proposed in this dissertation. It can enhance the wavelet video coding efficiency. In this algorithm, the selecting method of the hiberarchical reference frames set has considered fully about the temporal scalability of the video coding. It can support dynamic frame rate adjustment during the video transmission and thus provide the conditions for the dynamic bit rate transmission and control scheme.An improved algorithm based on the Embedded Zero Block Coding and Context Modeling algorithm, Bit Length Quad-tree No List Embedded Zero Block Coding and Context Modeling, is brought forward in this dissertation. This algorithm inherits the good merits such as the high coding efficiency and the good quality scalability from Embedded Zero Block Coding and Context Modeling algorithm. Simultaneously, it has solved the high memory consumption problem of Embedded Zero Block Coding. The wavelet video coding algorithm often consumes a lot of memory in the coding process, thus to some extent restricts its application. The improved algorithm proposed in this dissertation saves about 80% main memory consumption in the coding process in comparison with the original algorithm. It has the versatility in the wavelet coding algorithm domain and has the great value in the applications. A new stream packet method with stream description is brought forward in this dissertation by linking with the Bit Length Quad-tree Not List Embedded Zero Block Coding and Context Modeling algorithm. The encoder produces the stream description information in advance in the coding process. The stream description information can assist the media server to extract the sub-stream as need. The media server can extract the sub-stream and deliver it to the user according to the different bit-rate by using the stream description information as reference and simultaneously make the sub-stream close to the optimum according to the rate distortion criterion. Thus it can greatly reduce the operating complexity when the media server extracts the sub-stream. This method is valuable in the application of large-scale video transmission with large concurrent users.A dynamic bit rate video transmission and control scheme is brought forward in this dissertation. By the effective co-operation between the codec algorithm and the transmission control algorithm, the scheme greatly reduces the phenomena which seriously impairing the video exp eriences of users such as stop, buffering, interrupt, the mosaic and ghost in the pictures while video stream playing under the uncontrollable network. Thus it can improve the video experience of users. Based on the optimized wavelet video coding scheme proposed in this dissertation, the media server can use the dynamic bit rate video transmission and control algorithm proposed in this dissertation to extract sub-stream easily and deliver it to the users conveniently. Comparing with the scheme that the encoders are integrated directly in the media server to carry on the dynamic bit rate adjustment, the scheme proposed in this dissertation can greatly reduce the performance consumption of the media server. Thus it is suitable in the large-scale applications of video transmission.