A rate-distortion optimal hybrid scalable/multiple-description video codec

In this paper, we present a rate-distortion optimal hybrid scalable/ multiple-description video codec. Traditional scalable codecs produce bitstreams which can be partitioned into layers that form a hierarchy. Thus, in order for a particular layer to be useful to the decoder, all hierarchically higher layers also need to be available. Conversely, traditional multiple description codecs produce layers with no hierarchy. Thus, any of the layers can be decoded by itself and produce a video sequence of a certain quality. The more layers are available to the decoder, the better the corresponding video quality. The drawback of multiple description coding is that layers need to be correlated at the expense of compression efficiency. We propose a hybrid scalable/multiple description codec which produces a base layer and two multiple description enhancement layers. The base layer is required for decoding. If one or two of the multiple description enhancement layers are also received, the SNR of the received video sequence is improved. There is no hierarchy in the multiple description enhancement layers. The layers are constructed using a rate-distortion optimal partitioning of the DCT coefficients. Experimental results are presented and conclusions are drawn.