基于压缩感知的空间目标三维雷达成像方法

Three-dimensional imaging technique of space objects based on compressive sensing

对于稳定飞行的空间目标,俯仰向上的多圈次回波数据是稀疏分布的,从有限的观测数据中反演目标的三维反射率函数是不适定问题,观测噪声也会影响反演的结果,因此传统的FFT算法不再适用,必须引入适当的先验信息才能生成目标的三维图像。文章针对空间目标轨道的运动特性,首先推导了回波俯仰向表达式,然后结合目标散射中心稀疏分布特性和压缩感知原理,提出了一种基于多圈次稀疏观测的空间目标三维成像算法。该方法利用噪声单元估计噪声门限,当观测模型满足约束等距性质时,利用加权迭代的压缩感知算法进行成像处理,生成目标的三维图像。最后结合实测轨道模型,仿真验证了在低信噪比下,基于噪声估计的压缩感知算法能实现对目标三维像的精确重构。

For a space object at stable flight state,its multipass echoes data is sparsely distributed along the elevation.Reconstruction three-dimensional target reflectivity function from these limited observations is ill-posed.Moreover,noise will also degrade the reconstruction,so the ordinary Fourier transform is no longer applied,and the object can＇ t generate the three-dimensional unless the prior information can be acquired in advance.Aiming at the target orbit motion characteristics,the elevation expression is firstly derived in this paper,and then a three-dimensional imaging technique of space targets using multipass echoes is proposed combined with sparse distribution of target scattering centers and compressed sensing theory.The method employs noisy cells to estimate noise level and when measured model meets the restricted isometry property,the weighted compressed sensing iterative algorithm is used to produce the three-dimensional radar image of space object.Finally,combining with the real-time orbit model,the simulation results demonstrate that the compressed sensing based on noise estimation can achieve accurate three-dimensional image reconstruction in the low SNR.