Font Size: a A A

Rate Distortion Optimization For Wavelet Video Coding

Posted on:2007-10-30Degree:MasterType:Thesis
Country:ChinaCandidate:J XuFull Text:PDF
GTID:2178360242461954Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the development of computer technology, especially the development of network and applications based on network, more and more demands lead to emerging of new standards. International standards such as MPEG-4 and H.264 achieve outstanding performance. However, the hybrid scheme with DCT and block-based motion compensation used in these standards brings blocking artifacts into the reconstructed pictures and has difficulty in supporting scalability. Video codec using Discrete Wavelet Transform (DWT) does not have this kind of problem. Based on long term research in wavelet filter designing, image compression and video coding, the video coding framework based on DWT is improved, by support of the Project of National High Technique Research and Development (863). The techniques in this new framework are rate distortion optimized.The residual images in P frame have quite different statistics from images in I frame. Biorthogonal wavelet and multiwavelets are used to process these images respectively. Wavelet coefficients are quantized by SPIHT algorithm. Rate distortion optimization for new techniques is conducted in P frame, such as variable block-size motion estimation, multiple reference picture motion compensation, sub-pixel motion compensation. Motion compensation with variable block sizes, multiple reference pictures and sub-pixel can produce more accurate prediction frame and reduce the energy contained in the residual images. Using Lagrange function, codec selects the block sizes, reference frame and sub-pixel motion vector adaptively. Rate distortion based fast motion search algorithm decreases computation complexity without loss in quality. Overlapped block motion compensation (OBMC) can reduce the high frequency components in residual images, which benefits DWT. Adaptive block-size OBMC is a new way to conduct OBMC in variable block-size situation. Neighboring selection rules and different windows are tested in this new algorithm. For different elements, different entropy coders are used. Experiments on the wavelet video codec are conducted to evaluate and compare different techniques. Results suggest that rate distortion optimization improves the rate distortion performance of the codec. Optimized wavelet video codec has better rate distortion performance than MPEG-4.
Keywords/Search Tags:Video Coding, Wavelet Transform, Rate Distortion Optimization, Motion-compensated Prediction
PDF Full Text Request
Related items