The Research Of Entropy Coding In Video Compression

With the development of multimedia, network and communication technologies, video compression technology has become a hot spot. As one of the key technologies of video coding, entropy coding has being developed and improved.First, this paper introduces the video compression related concepts, mainly about the redundancy of video data and the development of entropy coding in video coding. On this basis, the paper focuses on improving CABAC on coding efficiency and coding speed. To improve the coding efficiency of CABAC, this paper proposes a new context model selection scheme for MVD coding. The new scheme uses the correlation between the horizontal and vertical components of MVDs instead of the correlation of the neighbor blocks to improve the probability estimation of symbols. The accurate probability estimation can therefore improve the efficiency of MVDs' coding with CABAC. The gain is 8.34% in P frames' mvd-y coding and 3.80% in B frames' mvd-y coding. To improve the coding speed, this paper focuses on improving the probability estimation update part of CABAC. The paper introduces the concept of packet and proposes to update probability estimation after every packet's coding instead of every symbol. Then at the analysis process of the improved algorithm, the paper introduces the concept of change frequency coefficient. Based on the concept of packed and change frequency, the paper proposes an adaptive high speed CABAC algorithm based on probability estimation update. It can improve the coding speed at 33.6% meanwhile the coding efficiency declines only 2.81% (QP=28).In the last part, the paper improves LZW algorithm for video compression and a variable length LZW coding algorithm with multiple dictionaries and compression ratio detection has been proposed. It has substantially improved compression efficiency.