基于复合正则化的稀疏SAR成像方法研究

Study on Sparse High Resolution SAR Imaging Method Based on Compound Regularization

随着高分辨率对地观测要求的不断提高,合成孔径雷达(Synthetic Aperture Radar,SAR)的应用将越来越广泛。针对高分辨率SAR成像存在数据量大、存储难度高、计算时间长等问题,目前常用的解决方法是在SAR成像模型中引入压缩感知(Compressed Sensing,CS)的方法降低采样率和数据量。通常使用单一的正则化作为约束条件,可以抑制点目标旁瓣,实现点目标特征增强,但是观测场景中可能存在多种目标类型,因此使用单一正则化约束难以满足多种特征增强的要求。本文提出了一种基于复合正则化的稀疏高分辨SAR成像方法,通过压缩感知降低数据量,并使用多种正则化的线性组合作为约束条件,增强观测场景中不同类型目标的特征,实现复杂场景中高分辨率对地观测的要求。该方法在稀疏SAR成像模型中引入非凸正则化和全变分(Total Variation,TV)正则化作为约束条件,减小稀疏重构误差、增强区域目标的特征,降低噪声对成像结果的影响,提高成像质量;采用改进的交替方向乘子法(Alternating Direction Method of Multipliers,ADMM)实现复合正则化约束的求解,减少计算时间、快速重构图像;使用方位距离解耦算子代替观测矩阵及其共轭转置,进一步降低计算复杂度。仿真和实测数据实验表明,本文所提算法可以对点目标和区域目标进行特征增强,减小计算复杂度,提高收敛性能,实现快速高分辨的图像重构。

With the continuous improvement of high-resolution earth observation requirements,the application of synthetic aperture radar(SAR)will become more extensive.Considering the issues with high-resolution SAR imaging,such as the need for large amounts of data,high storage difficulty,and long calculation time,the commonly used solution involves introducing compressed sensing(CS)into the SAR imaging model to reduce the sampling rate and reduce the amount of data.Using a single regularization as a constraint condition can suppress the side lobes of point targets and achieve point target feature enhancement.However,since multiple target types may be present in the observation scene,adhering to the multiple feature enhancement requirements using a single regularization constraint is challenging.Therefore,this study proposes a sparse,high-resolution SAR imaging method based on composite regularization.When using compressed sensing to reduce the number of calculations,the linear combination of multiple regularizations is used as a constraint condition to enhance the characteristics of different types of targets in the observation scene,and the requirements of high-resolution earth observation in complex scenes are achieved.The sparse SAR imaging method based on composite regularization introduces non-convex regularization and total variation(TV)regularization as constraints in the sparse SAR imaging model to reduce the sparse reconstruction error,enhance the characteristics of regional targets,reduce the influence of noise on the imaging results,and improve the imaging quality.The improved alternating direction method of multipliers(ADMM)is used to reduce the composite regularization constraint,which reduces the calculation time and reconstructs the image quickly.The azimuth-range decoupling operator is used to replace the observation matrix and its conjugate transpose to further reduce the computational complexity.The simulation and measured data experiments show that the proposed algorithm can enhance the features of point targets and regional targets,reduce computational complexity,improve convergence performance,and achieve fast and high-resolution image reconstruction.