Research Of Distributed Source Coding Using Improved Side Information

In wireless sensor networks,distributed video coding,hyperspectral image compression and other application scenarios,it is often faced to compress and to encode two or more correlated sources.Making full use of the correlation between sources to reduce redundancy and to encode efficiently is a task,that is need to be completed by distributed source coding scheme.Aiming at the shortcomings on the bit error rate and the compression ratio of the existing asymmetric distributed source coding schemes,a novel distributed source coding system based on improved side information was proposed in this paper,and two implementations are designed based on the different data to be coded.In this paper,the current popular distributed source coding schemes are summarized at first.Then,a codec scheme that can not only make full use of the correlation between sources,but also utilize the correlation among neighboring symbols as much as possible,to improve the compression ratio: Distributed Source Coding scheme Using improved Side Information was proposed.At the sender,the source output sequence is sampled and divided into a sampled and an un-sampled subsequences.The un-sampled sub-sequence is then compressed using a conventional arithmetic coder while the syndrome of the sampled sub-sequence is calculated.The receiver exploits the correlation between the side information sequence and the unsampled sub-sequence to estimate the sampled symbols,so that the correlation between the estimated sequence and the original sampled sub-sequence is improved,and finally,the syndromes of the sampled sub-sequence and the estimated sequence are used to jointly decode the sampled sub-sequence.For the two-dimensional plane data,the sampling pattern of Distributed Source Coding scheme Using improved Side Information was updated in this paper,in order to keep the correlation of the neighboring symbols as much as possible.At the same time,the interleavers are used to disperse the errors of each code block to reduce the Syndromes that needs to be transmitted.A variant of Distributed Source Coding scheme Using improved Side Information(two-dimensional)is designed.Further,the universal discrete denoising theory was extended to the research field of distributed source coding in this paper,and a side information improving algorithm based on the denoising theory was proposed to achieve the estimation of sampled symbols at the decoder.The experiment results in this paper show that compared with the distributed source coding schemes based on the Low-Density Parity-Check code and the arithmetic code,when the correlation among neighboring symbols is strong,the two proposed scheme can reach high compression ratio.The bit error rate of the reconstructed sequence can be kept relatively low when the correlation between sources is weak.They are efficient,practical distributed source coding schemes and are easy to be implemented.