Scalable video coding using allpass-based wavelet filters

Scalable video coding (SVC) allows to encode videos in a unique bit stream at several operating points corresponding to different quality, spatial resolution and temporal frame rate. In this paper, we apply the allpass-based wavelet filters to wavelet-based scalable video coding (WSVC) for improving the coding performance, in particular, at reduced spatial resolutions. The allpass-based wavelet filters used in this paper are orthogonal and symmetric, and have a better frequency response than the D-5/3 and D-9/7 biorthogonal wavelet filters. Therefore, some spatial aliasing effects can be suppressed, when the allpass-based wavelet filters are used as a spatial transform in WSVC. Moreover, the inverse MCTF transform is performed at full spatial resolution to avoid the imprecision of the deduced MVs. Finally, a coding performance comparison between the proposed allpass-based and conventional D-9/7 biorthogonal wavelet filters is shown in terms of coding efficiency.