Sparse Representation And Its Applicational Study In ISAR Imaging

Due to high resolution, inverse synthetic aperture radar(ISAR) has been widely used in the military and civilian fields. However, ISAR encounters many challenges such as high sampling rate, large data storage and difficult real-time processing. The investigation of these issues has become a hot research topic in the area of ISAR imaging.Based on the sparse characteristics of an ISAR target, further research on sparse representation in ISAR imaging has been conducted in this thesis. The CS-based ISAR model is established and sparse representations of stationary and maneuvering targets are studied. Furthermore, an algorithm using improved Fourier basis based on Bayesian compressed sensing for ISAR imaging is proposed for stationary target. For maneuvering target, a CS-based algorithm using adaptive chirp basis for ISAR imaging and a CS-based algorithm using time-frequency basis for ISAR imaging are proposed. The main work outlines as follows:1. For stationary target, an algorithm using improved Fourier basis based on Bayesian compressed sensing for ISAR imaging is proposed. Using the improved Fourier basis and the information of sparse prior, ISAR imaging based on Bayesian compressed sensing is achieved, simulation and real data verify the effectiveness of the proposed algorithm.2. For complex-motion target with angular acceleration, a CS-based algorithm using adaptive chirp basis for ISAR imaging is proposed. ISAR signal model verifies that the echo data is sparse in chirp basis. The proposal adopts DCFT algorithm and CLEAN technology to decompose the radar echo of every range cell into a series of chirp signals. Further, based on the frequencies and chirp rates of the decomposed chirp signals, the adaptive chirp basis of the radar echo of the corresponding range cell is derived and the radar echo is reconstructed by compressed sensing algorithm. Simulation and real data demonstrate the effectiveness of the proposed algorithm.3. For maneuvering target, this thesis studies the sparse representation of radar echo in frequency-time domain(1) Based on Gabor transform, a CS-based algorithm using Gabor basis for ISAR imaging is studied, and the experiment demonstrates that the algorithm is able to get clear ISAR image of the target with complex-motion by using very few data.(2) A CS-based algorithm using adaptive Wigner-Ville basis for ISAR imaging is proposed. Among the CS-based algorithms of Wigner-Ville basis, the cross-term can not be eliminated completely and thus the focusing effect can be affected. Therefore, DCFT algorithm and CLEAN technology are utilized to decompose the echo data of each range cell into several linear frequency modulation signals and the CS-based Wigner-Ville distribution of each corresponding linear frequency modulation signal is acquired. Through superimposing calculation of each separate time-frequency transform, Wigner-Ville distribution of the echo data of the corresponding range cell is derived. Simulation experiment verifies the effectiveness of the algorithm.