Research On Key Coding Technology In Distributed Video Coding System

With the rapid iterative update of image and video acquisition equipment technology and the rise of wireless network transmission schemes,applications such as mobile multimedia communications and wireless video sensor networks have begun to be widely popularized.These applications have limited computing resources,storage resources,and power reserves,and are not suitable for using traditional video encoding algorithms.Distributed video coding is an important application of distributed source coding,which has attracted widespread attention due to its low coding complexity,strong error resilience,and high scalability.After years of research and development,distributed video coding systems have made great progress in coding efficiency and rate-distortion performance,but this technology still needs to be studied from theory to practice.However,the coding efficiency of the Slepian-Wolf coding scheme,which is the most critical core in the distributed video coding system,still needs to be improved.Considering the scenarios of distributed video coding,a good Slepian-Wolf coding scheme should have the advantages of high encoding efficiency,low decoding error rate,and low encoding complexity at the same time.Based on the above reasons,this thesis chooses to design a new Slepian-Wolf coding scheme based on the most efficient source coding and lossless data compression method—arithmetic coding.The work and results obtained in this thesis are as follows:1.Source symbol Purging-based Distributed Conditional Arithmetic Coding(SPDCAC)is proposed,which is a new implementation idea of Slepian-Wolf coding based on arithmetic coding.This scheme purges a part of the symbols in the source sequence according to a fixed purging rate to reduce the number of actual coding symbols and improve the compression rate.The purged source symbols are not simply discarded but need to be used as the context of subsequent symbols.In this way,the correlation between adjacent symbols in the source sequence can be preserved,and the preserved correlation can also be used to assist in recovering those purged symbols in the decoding process.Compared with traditional distributed arithmetic coding,this coding scheme has higher compression efficiency,lower encoding complexity,and better decoding performance.In addition,according to the characteristics of the SPDCAC,an improved posterior probability calculation method is proposed in this thesis.This calculation method needs to adopt different calculation modes according to the occurrence of decoding ambiguity in the decoding process.The experimental results show that SPDCAC can effectively improve the decoding success rate after using this improved posterior probability calculation method.2.Distributed Conditional Arithmetic Coding based on Adaptive Source-symbol Purging(DCACASP)is proposed.This is also a scheme to obtain a higher compression rate by purging a part of the symbols from the source sequence.However,this scheme dynamically purges symbols according to the internal correlation of the source sequence,not based on a fixed purging rate anymore.For DACACSP,the higher the internal correlation of the source sequence,the higher the number of symbols is not to be encoded.The experimental results show that when the internal correlation of the source sequence is strong enough,this coding method can easily obtain a lower coding rate than SPDCAC,and the encoding complexity is also lower.A Maximum A Posterior-probability(MAP)decoding algorithm using multiple stacks is proposed.This algorithm explores the possible future decoding results of each decoding path by using an additional stack to allow any one of the decoding paths to continue to be tentatively extended forward.Different from the past MAP algorithm based on the decoding tree,when calculating the posterior probability of each decoding branch,this decoding algorithm no longer only considers the past decoding results but also incorporates the decoding results that may occur in the future into the cumulative posterior probability calculation results.The experimental results show that the decoding success rate is significantly improved after using the decoding algorithm in various Slepian-Wolf coding schemes based on arithmetic coding.3.A distributed video coding system using the source symbol purging scheme is proposed.In this system,both SPDCAC and DCACASP are used.In addition,this thesis adopts a different classification approach from past studies: the encoder not only classifies the pixel blocks in the WZ frame,but also classifies bitplanes.The encoder will flexibly select the appropriate Slepian-Wolf coding scheme and coding parameters to encode the data of the WZ frame according to the results obtained from the feedback channel.When the quality of the side information is good,the system will choose DCACASP to encode the bitplane to reduce the code rate;when the quality of the encoded information is not good enough,the system will choose the SPDCAC scheme and improve the decoding quality at the expense of a slight increase in the code rate.Simulation results show that the distributed video coding system using the source symbol purging scheme can achieve better rate-distortion performance than the classical DISCOVER system and DVC-based DVC system.Moreover,the system also achieves the goal of reducing the computational complexity of the encoding end of the DVC system by reducing the complexity of the Slepian-Wolf encoder.