H.263 scalable video coding and transmission at very low bitrates

In this dissertation fundamental issues of H.263 scalable video coding and transmission are investigated. In H.263 three forms of scalability are defined, SNR, temporal, and spatial. In this dissertation SNR and temporal scalability are addressed within the framework of low and very low bitrates culminating in a hybrid scalable layer. Furthermore, towards video transmission the issue of source and channel rate selections over scalable layers is solved optimally given the system characteristics.; Towards video coding, rate control algorithms are proposed for both SNR and temporal scalability. For temporal scalability a B frame selection mechanism incorporating motion, frame separation, and the bitrate is then presented. Based on the concept of service functions, this modified selection mechanism positions enhancement frames as warranted by the above criteria. Extending SNR and temporal scalability, a single enhancement layer merging the two forms of scalability is presented. A rate control for this hybrid layer is presented and the issue of selecting the type of scalability is also addressed. It is a novel method of type selection that selects between SNR and temporal scalability based on the source contents.; Finally, an optimal rate allocation algorithm for scalable video coding is presented. This algorithm utilizes rate-distortion theory to optimally select the source and channel rates over all scalable layers. This is an optimal selection in the operational sense given the system setup and the state of the channel. It utilizes the rate distortion characteristics of the source bitstream and the response of the channel to a class of channel coding techniques in the selection of the best source and channel rates for the channel.