Study On Bayesian Reconstruction Algorithm Based On Block Structure

Compressed sensing is a new data processing theory that integrates acquisition and compression.The main research contents include sparse processing,compression measurement,and reconstruction.The reconstruction algorithm is the core part of the compressed sensing algorithm.In recent years,the Bayesian reconstruction algorithm is the most prominent one.It uses a small amount of prior knowledge to obtain the posterior distribution through Bayesian theory and parameter optimization methods,and finally reconstruct the original signal information.Based on the traditional Bayesian reconstruction algorithm,this paper combines the inherent structural features of the signal with the Bayesian reconstruction algorithm,and reconstructs the signal obtained by a single sensor or multiple sensors,and then use the maximum expectation algorithm to optimize the parameters.The main research contents are as follows:Firstly,it introduces the basic knowledge,the development of compressed sensing,and several commonly used reconstruction algorithms.Then,the main contents of the Bayesian reconstruction algorithm,the main parameter estimation model and the application scope in Bayesian reconstruction are deeply studied.Secondly,Based on the analysis of the structural characteristics of the signal,the structural characteristics of the signal are introduced,especially the important role of block sparsity characteristics in the reconstruction algorithm.Then the signal with block structure obtained by the single sensor is reconstructed by the Bayesian method.The traditional block sparse Bayesian reconstruction algorithm is improved by the edge maximization method and iterative prior condition to reduce the parameters and iteration times.It can also reduce the computational complexity.The pipeline leakage signal was reconstructed with the improved reconstruction algorithm,and the experimental results and main influencing factors were analyzed at last.Finally,a multi-sensor joint sparse model was introduced.And the Bayesian reconstruction algorithm is applied to the multi-sensor joint sparse model to reconstruct the multitasking signal.Then reduce the dimension of the calculation process of the reconstruction algorithm to reduce the time required for reconstruction.At last,the reconstruction accuracy and reconstruction performance of this method and other traditional reconstruction algorithms are compared by experiments.And the experimental results show that this method works well.