压缩感知机动目标ISAR成像新方法

New compressed sensing algorithm for ISAR imaging of maneuvering target

基于Fourier基的压缩感知(Compressed Sensing,CS)算法已被成功应用于平稳运动目标的逆合成孔径雷达(Inversed Synthetic Aperture Radar,ISAR)成像。但由于建模时对ISAR回波方位相位高次项的忽略,Fourier基矩阵对机动目标回波数据方位信息的稀疏表示失效,导致对机动目标的成像在方位向模糊。鉴于时频分析技术良好的时频局部化特性,将其引入到雷达回波方位向分析中,以改进用于表示雷达回波数据的稀疏基,实现对选定时间切片内回波数据多普勒频率的稀疏表示。改进后的基矩阵在通过CS技术解析回波在时间切片内方位信息的同时,又保证了利用有限数据成像的分辨率。与基于Fourier基CS成像等现有方法相比较,新方法在方位向的成像质量上有较大改进。仿真实验验证了算法的有效性。

The Fourier basis compressed sensing（CS） algorithm was applied in inversed synthetic aperture radar （ISAR） imaging of smoothly moving target successfully. But it ignored the higher order terms of ISAR echo in azimuth when constructing the ISAR echo model, the sparse representation based on Fourier basis for azimuth information of ISAR echo whose target is maneuvering was invalid, which leaded to the lack of information of Doppler frequency in the local range of time domain. As a result, the imaging results of maneuvering target were blurred in azimuth. The time-frequency analysis technology, because of its good time-frequency characteristics in local, was introduced into the analysis of ISAR echo in azimuth： A Gauss window was used to improve the sparse basis which represented the Doppler frequency of ISAR echo data in a selected short time slice while the size of sparse basis matrix remained constant. Then CS technology which includes time-frequency analysis based sparse basis to represent the echo data, Gauss random observation matrix to reduce the sampling rate and Orthogonal Matching Pursuit （OMP） algorithm to solve the coefficients of the sparse representation was used to analyze the azimuth information in that time slice. As a result of CS＇s advantage in improving resolution,the resolution of image supported by limited data was high at the same time. The target models of uniformly accelerated motion and varying accelerated motion were both designed to simulate the ISAR echo data of maneuvering target. Compared with the existing imaging methods such as Fourier basis CS algorithm, Range Doppler（RD） algorithm and the Range instantaneous Doppler（RID） algorithm based on Gabor transform, the new one achieves significant improvement in terms of imaging results in azimuth. The corresponding imaging results show the effectiveness of the algorithm from both subjective observation results and objective evaluation indicators including Peak Side Lobe Ratio（PSLR） and azimuth resolution.