Research On The Video Coding System Based On The Wavelet Morphology

Recently,many applied image coding algorithms are mixed coding methods based on the spatial motion estimation and compensation,the DCT and quantification of displaced frame difference and entropy coding. But the algorithms based on DCT have their inherent disadvantage that is "blocking effect" and "mosquito noise". The quality of the restored images is much worse at low bit rate, so the algorithms isn't suitable for the situation of low band. Having the good characteristics of temporal-frequency localization and so on,wavelet transform is very suitable for describing nonstationary signal . And because the significant coefficients within each subband have clustering characteristics after wavelet transform, we use the morphological dilation operator to predict, extract and code the significant coefficients within each subband. It overcome the shortcoming of wavelet coding algorithms, decreased the computing complexity and achieved better performance than the encoding in EZW. In this thesis,we emphatically discusses the theory of the video coding system based on the wavelet field and morphology following the basic theory of video coding. We described and analyzed the classical wavelet coding algorithms and their problem. Then we propose a novel video-coding ameliorative algorithm based on wavelet and morphology. video coding system consists of I-frame coding and P-frame coding. As to I-frame compression, we add sense quantization, propose a ameliorative wavelet coding algorithms based on EZW and MRWD, and achieve arithmetic coding through pointing mode. As to P-frame compression, block is only transformed by wavelet. Then low frequency subband is performed motion estimation and estimated motion vector generalize to high frequency subband. We use full search and biliner interpolation of half-pixle accuracy in motion estimation. Meanwhile, the residual figure in motion compensation is encoded by wavelet. The stimulation results indicate that the performance of the video-coding algorithm based on wavelet and morphology are superior to that based on spatial field not only in coding efficiency but also in subjective quality of the decoded image.