基于快速阈值迭代的SAR层析成像处理方法

SAR tomography method based on fast iterative shrinkage-thresholding

合成孔径雷达层析成像(synthetic aperture radar tomography,TomoSAR)是将合成孔径原理应用到高程向进行三维成像,相比于传统的二维成像,增加了高程向的信息。传统谱估计方法可用于SAR层析成像,但其高程向分辨率较低。对于高程向分布稀疏的场景,压缩感知(compressive sensing,CS)方法可以用于高程向重建,且具有超分辨能力。阈值迭代算法(iterative shrinkage-thresholding,IST)可用于SAR层析成像,但其收敛速度比较慢。介绍了一种快速阈值迭代算法(fast iterative shrinkage-thresholding,FIST)用于SAR层析成像,该方法不仅保持了IST算法计算的准确性,而且具有较快的收敛速度。本文通过仿真实验说明FIST算法在多散射体分辨、单散射体位置估计等方面的特性,并利用TerraSAR-X北京地区实际数据进行SAR层析成像,分析成像效果。研究结果表明FIST算法在多散射体分辨、单散射体位置估计方面优势明显,其应用于SAR层析成像具有较好的成像效果。

synthetic aperture radar tomography （TomoSAR） applies the synthetic aperture principle into the three-dimensional imaging of the elevation direction. Compared with the traditional two-dimensional imaging, it increases the elevation information. Traditional spectrum estimation methods can be used for SAR tomography, but the elevation resolution is lower. For the sparse scenes in the elevation direction, compressive sensing （CS） methods can be used to reconstruct the elevation information, and have the super-resolution capability. The iterative shrinkage-thresh- oldingCIST） algorithm can be used for TomoSAR , but its convergence rate is slower. A fast iterative shrinkage-thresh-olding （FIST） algorithm is introduced for TomoSAR , which will not only maintain the accuracy of the 1ST algorithm, but also has a faster convergence rate. The characteristics of the FIST algorithm are shown in multiple scatterers dis-tinguish, single-scatter position estimation and other aspects through simulation experiments. Then, the TerraSAR-X actual data of Beijing region is used to analyze the imaging result of TomoSAR. The research result shows that the FIST algorithm has obvious advantages in the aspects of multiple scatterers distinguish and single-scatter position esti-mation. It can apply to TomoSAR with better imaging results.