基于门限正交匹配追踪（TOMP）的线阵SAR三维成像方法

摘要

In this paper, a new 3D linear array SAR imaging method based on threshold orthogonal matching pursuit (TOMP) method was proposed. Because of the imaging theory of conventional method was based on coherent integrating, the side-lobe of the image was high and the main-lobe was wide. So the resolution of the image reconstructed by conventional method was limited to the upper bound which was called Rayleigh Limitation. Some scholars had proposed a 3D imaging method based on OMP which can break the Rayleigh Limitation and reduce the data size of SAR echos. The OMP method had an shortcoming which was need to know the sparsity of the observation area, but the sparsity of the observation area was always unknown in SAR application fields. In order to apply OMP method, the sparsity was usually set to a larger value than the true sparsity of the observation method which will lead to side-lobe appear in the image reconstructed by OMP and reduce the efficiency of OMP. The proposed method in this paper which was called TOMP, can overcome the shortcoming of the OMP. Compared with OMP, the image quality obtained by TOMP was better and the efficiency of TOMP was also higher. Simulations was given to prove the superiority fo TOMP.%在本文中提出了一种基于门限正交匹配追踪（TOMP）算法的线性阵列 SAR 三维成像超分辨方法。由于传统阵列SAR三维成像方法（如：BP、RD）都是基于相参积累原理的，其缺点是峰值旁瓣比较高，主瓣宽度较宽，其分辨率受限于瑞利限，雷达图像的分辨率较低。有学者提出了基于正交匹配追踪（OMP）的阵列SAR三维成像算法，其分辨率高于传统三维 SAR 成像方法的理论分辨率且进行稀疏采样大大降低了雷达数据的规模，但是OMP 算法有个很大的缺点就是要设定观测场景的稀疏度。在雷达领域中这个观测场景的稀疏度往往是未知的，因此为了能够准确重构出观测场景，OMP 算法设定的稀疏度的值通常大于信号的真实稀疏度，这将导致OMP算法重构得到的雷达图像中出现旁瓣，图像质量下降，算法效率降低。在OMP算法的基础上，本文提出了门限正交匹配追踪（TOMP）算法，该算法克服了OMP算法需要预设场景稀疏度而导致的缺点，得到的雷达图像的质量更好，算法效率更高。在文章最后给出了仿真实验证明了TOMP算法相对于OMP算法优势。