A Study on Image Reconfiguration Algorithm of Compressed Sensing

Abstract

Compressed sensing theory is a subversion of the traditional theory. The theory obtains data sampling points while achieves data compression. The main content of this thesis is reconstruction algorithm. It’s the key of the compressed sensing theory, which directly determines the quality of reconstructed signal, reconstruction speed and application effect. In this paper, we have studied the theory of compressed sensing and the existing reconstruction algorithms, then choosing three algorithms (OMP, CoSaMP, StOMP) as the research. On the basis of summarizing the existing algorithms and models, we analyze the results such as PSNR, relative error, matching ratio and running time of them from image signal respectively. In the three reconstruction algorithms, OMP algorithm has the best accuracy for image reconstruction. The convergence speed of CoSaMP algorithm is faster than that of the OMP algorithm’s, but it depends on sparsity K quietly. StOMP algorithm on image reconstruction effect is the best, and the convergence speed is also the fastest.