Research Of Thresholding Iterative Algorithm In Compressed Sensing

The transmission and reconstruction of a large amount of information challengecommunication technology in society nowadays. Then a novel information theory namedCompressed Sensing has been proposed by researchers in recent years. It points out: a highdimension signal with sparsity and compressibility can be projected to a low dimensionmeasurement vector which later can be used to reconstruct signal. Therefore, thereconstruction algorithm based on Compressed Sensing is significant for improving thetechnology of signal recovery at communication terminal.This paper mainly proposes a new reconstruction method-thresholding iterative algorithm.Innovation points are as follows:(1) Thresholding iterative reconstruction algorithm based on affine. Firstly, I propose a newthreshold function to improve conventional iterative algorithm, extract the key informationinside signals to a large extent. By comparing to the result from conventional method, thethresholding iterative algorithm has improved the accuracy of construction greatly(2) Thresholding iterative reconstruction algorithm based on regularized reweight. This partrefines the method of computing reweight, accelerates the rate of convergence and approachesto the optimal solution approximately. The method with regularized reweight can get a smallererror than the method with affine. Also, the threshold function has a good effect on thereconstruction.(3) The method of adaptively choosing threshold. Firstly, I propose a function that describes arelation between an independent variable and threshold. Therefore, with algorithm running,there must be an error under the standard error which stands for accurate reconstruction. Andthe corresponding value of threshold is the desired threshold. The advantage of this method isto save computation time and enhance the efficiency of algorithm. The experiment shows thatthe threshold chosen adaptively is approximately equal to the empirical one and certifies theaccuracy of the method of adaptively choosing threshold.