S. Van Leuven, K. De Wolf, P. Lambert, and R. Van de Walle (Belgium)
Scalable Video Coding, spatial scalability, macro block types, fast mode decision
Recently, the scalable video coding extension of H.264/AVC has been finalized. This extension, commonly referred to as SVC, allows refinement layers to be added in the bit stream. These layers can increase the visual quality (quality scalability), frame rate (temporal scalability) or resolution (spatial scalability). During encoding, the layered design results in an increased computational complexity, partly due to the motion estimation step that has to be performed for each spatial layer. Fast mode decision algorithms can be used in the enhancement layer to reduce the complexity, while this will result in a degraded rate-distortion performance, i.e., a lower visual quality and/or a higher bandwidth will be obtained. Since there has not been a profound analysis of the macroblock types in the enhancement layer compared to the used macroblock types of the co-located macroblock in the base layer, this paper shows the correlations between the macroblock type in enhancement layer and the macroblock type of the co-located macroblock in the base layer. Furthermore, the analysis takes the quantisation and resolution of both layers into account. These results are used as a theoretical background for a proposed fast mode decision model.
Important Links:
Go Back
