We consider the problem of scalable video coding and its application to video transmission over wireless channels. A scalable video encoder produces a bit stream which can be partitioned into layers which form a hierarchy. The base layer must always be transmitted. One or more enhancement layers can be transmitted to improve the quality of the received video sequence. Video scalability is particularly useful in transmission over heterogeneous networks and transmission over lossy channels using unequal error protection.; The three most common types of scalability are Signal-to-Noise Ratio (SNR), temporal and spatial. We propose a new method for SNR scalability which is based on the partitioning of the Discrete Cosine Transform (DCT) coefficients into a base layer and one or more enhancement layers. Depending on bandwidth availability, one or more layers can be transmitted to provide progressively higher SNR of the received sequence. The partitioning of the DCT coefficients is done in a way that is optimal in the Operational Rate-Distortion sense. The partitioning algorithm utilizes Lagrangian optimization and Dynamic Programming.; We also consider the problem of optimal bit allocation between source and channel coding for video transmission over wireless Rayleigh fading channels. The proposed scalable encoder is exploited to allow for different levels of protection for each scalable layer (unequal error protection). An algorithm is presented for the optimal allocation of an available bit budget across scalable layers and, within each layer, between source and channel coding.
展开▼
