| In recent years, Compressive sensing theory is one of the hottest topics in signal processing area. The theory asserts that one can recover certain signal and image from far fewer samples or measurements than traditional methods. In many applications, people would like to sample fewer samples or have to sample uncompleted information in some case. And compressive sensing theory can solve this problem exactly. Till now, compressive sensing theory has already applied to many practical applications, such as medical imaging, radar imaging, communication, etc.In the framework of compressive sensing theory, there are three cores:sparse representation, measurement matrix and reconstruct algorithm. How to design the measurement matrix to reduce the samples and ensure the good quality of reconstruction? How to select a proper reconstruct algorithm to improve the quality of reconstruction? How to make the compressive sensing system faster and efficient? These problems are the key problems of compressive sensing in image processing area. It is meaningful to solve these problems. This article will focus on the design of measurement matrix and the selection of reconstruct algorithm. The fellow work will be shown in the content of this article.(1)Generally, the design of measurement matrix is based on the matrix that has random property, like random Gaussian matrix. But this kind of matrix will cause many problems in the measurement process of compressive sensing. To make the measurement process fast and efficient, a new compressive sensing scheme using the structured random matrix (SRM) and discrete periodic Radon transform (DPRT) is proposed.(2)In the part of reconstruct algorithm, we will analyses different kinds of reconstruct algorithms and compare their reconstruction performance. To improve the performance of compressive sensing system, we will introduce some techniques on selecting the proper reconstruct algorithm based on different applications. |
PDF Full Text Request