Enhanced interframe wavelet video coding considering the interrelation of spatio-temporal transform and motion compensation

摘要

High coding performance and temporal, spatial, and SNR scalability are key elements required in future video applications. Within hybrid video coding schemes, scalability cannot be efficiently achieved due to the recursive coder structure. In motion-compensated interframe wavelet video coding, temporal filtering using wavelet filters and spatial wavelet decomposition are successively combined in an arbitrary manner, allowing for efficient spatio-temporal decorrelation and full scalability. Depending on the particular succession of spatio-temporal decomposition, typical artifacts are induced, resulting in reduced subjective and objective coding performance. In this paper, these problems are addressed within the scope of a t+2D scheme, where temporal filtering is followed by spatial transform coding, and possible solutions are discussed. Techniques for deblocking, adaptation of temporal and spatial filters, as well as for optimization of rate allocation are presented.