Quality control on the fully scalable wavelet video coder

The wavelet-based video codec is capable of providing scalable bitstreams such that quality of service (QoS) among heterogeneous networks can be manipulated in a unified framework. The three-dimensional wavelet video coder 3D WVC had been developed to provide this fully scalable bitstream. However, the quality control or rate allocation is performed on the scale of one GOP such that the reconstructed pictures suffer temporal quality fluctuations. The theoretical coding model provided ratio parameters among subbands to smooth temporal picture quality. However, it needs further adaptation for dealing with practical WVC. We propose to investigate the rate-distortion relation, between the decomposed subbands (Ψ) and the reconstructed pictures in one GOP (P), under which the rate allocation for constant quality (RACQ) can be operated. Experiments show that the temporal quality fluctuations can be reduced to 10-30 times smaller as compared to 3D-SPIHT and 5-10 times smaller to the ratio parameter approach.