Research and Implementation of Dwt-based Video Codec System at Low Bit Rates
|School||Huazhong University of Science and Technology|
|Keywords||Discrete Wavelet Transform Significant test Backward prediction Refinement of mathematical morphology|
In the video or image coding , discrete wavelet transform (DWT) , compared to the discrete cosine transform (DCT) with scalability, no blocking effect , and are consistent with the advantage of the human visual system . To this end , the latest video coding standard H.264 modeled , the analysis of its principle , structure and key technologies , elaborates JPEG2000 is how to use wavelet transform coding , and as a basis for a low DWT-based video codec bit rate . By a first-order two - dimensional discrete wavelet decomposition, the separated high and low frequency components of each frame of the video series are processed separately. The low frequency part of the H.264 coding compression; high frequency portion is divided into the I - and P-frames are encoded , based on the characteristics of the wavelet coefficients , P frames with backward prediction techniques to predict , and using mathematical morphology Refinement method correction , cyclical the transmission of I frame data after the passage to the inter - frame motion vector prediction and motion compensation , the high - frequency part of the P frame to strengthen and reduce the P frame error accumulation ; the other hand , the I frame only encoding the significant coefficient of transmission in a specific proportion superset of significant coefficients, and using the adaptive scan mode , the run-length coding and variable length coding of these ultra significant coefficient encoding compression . In addition , in-depth research and analysis of the key modules in the system to achieve specific performance . Optimization of the wavelet basis selected , high-frequency significant coefficient backward forecasting , mathematical morphology thinning algorithm , entropy encoding , and the entire system of coding control and other issues specific analysis and simulation . Experimental results show that at low bit rates , the overall performance of the video coding scheme is better than H.264 , especially I - frame encoding , and its performance is significantly better . Strong texture detail and the quality of the reconstructed video retention capacity , the outline of the image is clearer , subjective visual quality has a larger increase .