Research On Compressed Sensing Restructing Algorithm Based On Interior Point Method

As one of the imoprtant research field, the compressive sensing theory was widely put into use on information collection, medical images and remote sensing image and so on. Compressive sensing had a series of advantages, such as less information collection and more efficient, more importantly than all of that,it could work with less accurate on hardware system. Thus it was extensively attentioned when once put forward and acquired a large number of scientific research achievements. With the development of the theory, the computational efficiency and time-consuming of the restructing algorthm were optimized. In this paper a major theme in research is the image reconstruction algorithm based on interior point method, divided into the following three parts:(1) Details concerning the basis factors of the compressive sensing, the implementation of the mainstream algorithm, Systematic learning the main problems in the study. By establishing the mathematical model to analyze the differences on different algorithm, draw a conclusion that under the study of compressive sensing it could incresse efficiency on calculating.(2) Study about the orthogonal matching pursuit based on greed iterative algorithm and the basis pursuit based on convex optimization. By comparing the mathematical model and realization method to come to the conclusion that: the computing time complexity of the OMP is high and long operation time; The BP has a high computation complexity but the IPM based on BP has a lot of room for improvement, and then give the implement method of the IPM. Through experiment to conclude that the implement method of the IPM has advantages on run time and restoration efficiency.(3) Against the disadvantges of the IPM, specially improved algorithm is put forward. Based on the IPM, introduce of preconditioning conjugate gradient operator to replace the direct operator or conjugate gradient operator. In this way, it can improve the convergence speed of the algorithm, reduce the computational complexity and reduce the running time. It through a series of experiments show that the new algorithm reduced the one order of magnitude and reduced the runtime.