Compressed Sensing Based Staring Imaging For Large Scene

The traditional staring imaging radar generally utilized real aperture radar imaging techniques.The resolution of real aperture radar imaging is limited by the antenna aperture,which affects the development of traditional staring imaging radar.In recent years,the application of compressive sensing(CS)theory in radar imaging has broken through resolution limitation of the antenna aperture.However,CS based staring imaging radar suffers huge computational and memory burden in the case of the large scene,as the dimension of the sensing matrix is very large.The main contributions of this thesis are summarized as follows:1.The computational complexity is large in 1-dimensional received signal model when we reconstruct large scene.To solve this problem,we investigating the two-dimension received signal model for staring imaging radar.Firstly,for the range-azimuth two-dimensional staring imaging MIMO radar systems with transmit-receive linear array,when orthogonal waveforms are transmitted,two-dimensional received signal model is obtained directly after the matched filter.The transmit steering matrix and receive steering matrix are separated in this model.Secondly,for the range-azimuth-elevation three-dimensional staring imaging systems,the transmit antenna is planar array.We can factorize the large-dimensional steering matrix of the planar array into two small-dimensional matrice by exploiting the special element structure of the planar array.The received signal model transformed to a twodimensional received signal model.The two-dimensional received signal model is different from the 1-dimensional one.The receive data of the former is a matrix and the later is a vector which comes from the vecteration of the matrix.This results to the reduced dimension of the matrix operations in CS algorithm.2.The CS algorithms are studied according to two-dimensional receiver signal model.Firstly,two kinds of reconstruction performance of traditional greedy algorithms in compressive sensing are introduced,and the detailed flows of these two classical greedy algorithms are summarized.Secondly,a modified Stagewise Orthogonal Matching Pursuit(St OMP)algorithm is proposed according to the two-dimensional received signal model based on compressive sensing.Theoretical analyses show that the modified algorithm reduces the computational and memory burden compared with St OMP.Numerical simulation results demonstrate the effectiveness of the proposed algorithm.Then the modified CS algorithm is applied to the two-dimensional received signal model based staring imaging radar.The simulation results show that the modified algorithm has obvious advantages when reconstructing the large scene.In the end,a graphical interface of radar staring imaging system based compressive sensing is developed,which makes it convenient to evaluate the performance of the algorithms.