Allocating low bit-budgets to intra-coded pictures can often lead to high quantisation and loss of important high frequency information. Coding at lower resolutions can alleviate some of this distortion when utilising effective resampling techniques and filters that minimise the additional distortion introduced as a result of spatial resampling. However, it is not always possible to reconstruct each portion of a picture to the same or similar level of quality; within natural scenes frequency content tends to vary spatially and therefore a fixed scale factor may either cause loss of high frequencies or oversampling of low frequency content. SHVC provides many of the tools required for enhancing rate-distortion performance of intra-coded pictures but it is not used for this purpose. In this paper we analyse the performance of a mixed resolution coding approach by using inter-layer prediction within SHVC to construct intra-coded pictures as a combination of different spatial resolutions. We evaluate the performance of coding with and without an enhancement layer for varying bit-budgets and compare these results to standard HEVC to provide insight regarding the optimisation of intra-coded pictures.
|Publication status||Accepted/In press - 2016|