基于近似观测和最小熵约束的SAR稀疏自聚焦方法

当采样率较低以及重建散射点数量较多时, 基于两步优化的稀疏自聚焦方法收敛速度较慢且容易陷入误差较大的局部最优解, 导致自聚焦失败。针对此问题，提出了一种基于近似观测和最小熵约束的稀疏自聚焦方法。首先, 为解决测量矩阵规模大、内存占用高的问题, 构建了一种基于近似观测的稀疏自聚焦模型, 在聚焦图像的傅里叶变换域引入误差相位。然后, 在采用最大似然估计器估计误差相位时增加了最小熵约束, 同时采用相位梯度自聚焦法提供误差相位的初始解, 有效降低了迭代次数并使迭代结果更接近全局最优解。机载合成孔径雷达的实测数据成像结果表明, 与常规自聚焦方法相比, 所提方法具有更快的收敛速度和更稳定的自聚焦性能。

SAR sparse autofocusing method based on approximate observation and minimum entropy constraint

When the sampling rate is low and the number of reconstruction scatterers is large, the convergence speed of the two-step optimization based sparse autofocus metgod is slow and it is also easy to fall into the local optimal solution with large reconstruction error, causing the autofocus to fail. Aiming at this problem, a sparse autofocus method based on the approximate observation and the minimum entropy constraint is proposed. First, to solve the problem of large-size measurement matrix and high memory footprint, a sparsity-driven autofocusing model based on approximate observation is constructed, in which the phase error is introduced in the Fourier transform domain of the focused image.Then, the minimum entropy constraint is added into the maximum likelihood estimation of the phase error. Besides, the phase gradient autofocus method is used to provide the initial solution for the phase error, which effectively reduces the number of iterations and make the iterative result close to the global optimal solution. The imaging results of airborne synthetic aperture radar(SAR) data show that the proposed method has faster convergence speed and more stable self focusing performance than the conventional autofocusing method.