Scalable video coding: Analysis and coding performance of inter-layer prediction

The emerging scalable video coding specification, which is a layered extension of the H.264/AVC specification, introduces several inter-layer prediction modes to increase the compression rate. A major drawback of these prediction modes, is the increased complexity of the encoder. In this paper, we focus on the coding performance using inter-layer prediction modes in spatial enhancement layers. From our performance results, we see that enabling the inter-layer intra prediction mode results in bit rate savings of 9.9% on average for the tested sequences and the combinations of quantization parameters without significantly affecting the PSNR (a slight increase of 0.2% on average). The downside of this prediction method, is an increased encoding time of 18 to 22%. For interpredicted pictures, the increase of the coding time is significant: 133.32% on average, with an observed maximum of 259.95%. The highest bit rate reductions (up to 27.47%) and increase in PSNR (0.21% on average) are observed when the quantization of the enhancement layer is high.