Spatial prediction in scalable video coding

Spatial scalability allows the transmission of images at different levels of resolution in a single bit stream, offering potential spectrum savings with respect to simulcast transmission. The authors consider a two-layer MPEG-2 system which uses spatial scalability to transmit a conventional television signal as a base layer, and an HDTV signal as an enhancement layer or alternatively a conventional television signal as the enhancement and a SIF interlaced signal as the base layer. The issue of interlaced to interlaced conversion, and its effect on the overall performance are studied. The usefulness of the base layer signal in the prediction of the upper layer signal is discussed. Finally, the performance of spatial scalability is compared with a simulcast approach for different bit rates at the lower layer, while keeping the overall bit rate constant