Research On Distributed Video Encoding And Decoding Based On Compressed Sensing

Distributed Video Coding(Distributed Video Coding,DVC)is a video coding method of independent coding and joint decoding.It transfers computationally complex modules from the encoder to the decoder,which reduces the coding complexity of the encoding end and is widely used in wireless video communication and video.Various scenarios such as sensor networks and mobile digital cameras.Compressed sensing(CS)uses the sparsity of the signal to restore the entire desired signal from fewer measured values,and is used in signal processing and other fields.At present,there are some achievements in the research of distributed video(DCVS)coding and decoding based on compressed sensing.However,for the existing side information algorithms,there are always points to be improved in the complexity of the side information generation algorithm or the quality of the generated side information needs to be improved,which affects the decoding effect.As for the existing signal reconstruction algorithms,most of them are the optimization and improvement of the classic algorithms.The research on the application of the new algorithm to improve the reconstruction quality is still very little.The research on the application of the new algorithm needs to be proposed.In order to improve the existing problems of the above-mentioned distributed video compressed sensing system codec,this paper conducts a research on distributed video codec based on compressed sensing,proposes an edge information generation algorithm,and applies the algorithm to optimize signal reconstruction to achieve the improvement of the system The goal of the video decoding effect.First,determine the basic system framework of the CS-based DVC framework studied in this paper,and focus on the measurement matrix and sampling rate in the framework.The influence of Gaussian random matrix,chaotic random matrix and partial Hadamard matrix on the peak signal-to-noise ratio of system decoding at different sampling rates is studied and discussed.Secondly,under the basic frame structure,this paper makes full use of the similarity between the two key frames and non-key frames,uses the advantages of the frame interpolation side information generation algorithm,and combines the K-SVD algorithm with the initial side information generation algorithm for motion estimation.The sparse dictionary learning edge information algorithm is optimized,and an improved sparse edge information generation method,motion estimation double mean dictionary learning edge information generation optimization algorithm,is proposed.Finally,this paper analyzes the algorithm principle of the traditional sparse signal reconstruction algorithm MP matching tracking algorithm and its typical optimization algorithm OMP orthogonal matching tracking algorithm and CoSaMP compressed sensing matching tracking algorithm.According to the theoretical basis of Bayesian decision and Bayesian learning,the signal reconstruction part of multi-vector sparse Bayesian learning algorithm applied in distributed video coding system is proposed as a signal reconstruction algorithm.This paper uses sampling rate,calculation time,peak signal-to-noise ratio of video signal frame reconstruction as the main parameters,and performs simulation experiments with the comparison algorithm.Experiments show that the motion estimation double mean dictionary learning edge information generation algorithm proposed in this paper has low computing time and relatively high peak signal-to-noise ratio data,and the multi-vector sparse Bayesian learning algorithm is used as a reconstruction algorithm to decode distributed video The peak signal-to-noise ratio has a certain improvement.Therefore,the edge information generation algorithm proposed in this paper and the applied signal reconstruction algorithm have an optimized effect on the video decoding of the system.