Study Of ISAR Image Enhancement Based On Compressed Sensing And Regularization

Inverse synthetic aperture radar(ISAR) can obtain high resolution images of non-cooperative targets and has found wide applications in military and civil areas. With increasing surveillance range of ISAR and targets' maneuvering motion, the signal-to-noise ratios(SNR) of ISAR images are generally low. To better extract targets' features, i.e. the geometry, it is necessary to study the ISAR image enhancement technique.ISAR image enhancement emphasizes improving the visual effect, and facilitates subsequent image segmentation and target recognition. This thesis first reviews the recent development of ISAR imaging and image enhancement techniques, and then presents the basic principles of ISAR imaging and corresponding simulation results. Next, utilizing the sparsity of ISAR images, compressed sensing(CS)-based ISAR image enhancement is discussed in detail. After that, the regularization method is proposed to enhance the ISAR images and the experimental results of real data are also given. Finally, the thesis is concluded. The main content of this thesis is summarized as follows.(1) Chapter 1 introduces the research background of this thesis, as well as the recent development of ISAR imaging and radar image enhancement techniques.(2) Chapter 2 presents the basic principles of ISAR imaging. Firstly, the imaging geometry and signal model are studied, including the theoretical expressions of range and cross-range resolution. Then, it is shown that the target's movement can be divided into the translational motion and rotational motion, and hence translational motion must be compensated to achieve two dimensional imaging. After that, the generation of migration through range cell(MTRC) is discussed and Keystone transform is utilized to compensate the MTRC. Finally, simulation results illustrate the ISAR imaging and effectiveness of Keystone transform.(3) Chapter 3 focuses on the CS-based ISAR image enhancement, as well as the sparse expression, incoherent measurement and corresponding reconstruction algorithms. Firstly, the basis pursuit method(BP) and orthogonal match pursuit(OMP) method are discussed to reconstruct the sparse signal. Then, CS is utilized to improve the cross-range resolution and enhance the ISAR image. Well focused images can be obtained based on CS under sparse ISAR echoes or short coherent processing interval(CPI). Because of the sensitivity to the noise, an improved CS(ICS) method with modified weight matrix is introduced and discussed. Finally, the simulation results verify the effectiveness of the introduced CS and ICS.(4) Chapter 4 emphasizes the regularization–based ISAR image enhancement technique. Similar to CS, the regularization method utilizes the sparsity of ISAR images and converts the ISAR image enhancement problem to one optimization problem. Meanwhile, the edge information is sufficiently considered in regularization method. This chapter discusses the image enhancement methods in frequency domain, complex image domain and gray image domain, respectively. Finally, to determine the coefficients of different regularization terms, several experiments are done on real data and the corresponding experienced values are given. The experimental results show the effectiveness of the proposed methods.(5) Finally, chapter 5 concludes this thesis and presents the future research focus.