| Summary: | 碩士 === 國立臺灣科技大學 === 電機工程系 === 93 === The wavelets demonstrate better compression capabilities than DCT. In addition, the wavelet-based codestreams are highly scalable which cannot be accomplished easily by block-DCT coding approach. This scalability for coding and transmitting two-dimensional signals, e.g., images, can be extended to three-dimension to deal with videos. However, it suffers temporal picture quality fluctuations when using wavelet video codec (WVC) frameworks. The error propagation is unavoidable in WVC. The theoretical coding control parameters for smoothing temporal quality fluctuations can be derived with high-bit-rate coding assumptions. In deal with practical WVC codec, the signal properties may not follow high-bit-rate coding behavior such that it needs further adaptation to deal with the practical video coding. We proposed to investigate the rate-distortion relation, between the decomposed subbands and the reconstructed pictures in one GOP, under which the rate allocation for constant quality (RACQ) can be operated. The proposed rate allocation method is carried out after the initial rate allocations by theoretical parameters are performed. Simulations show that the temporal quality fluctuations can be largely reduced as compared to the theoretical one and the 3D-SPIHT. Motion compensation can also be used before performing the RACQ.
|
|---|
