基于最小全变差的SAR图像自聚焦算法

该文分析了方位一维图像的全变差范数随二次相位误差的变化情况，得到全变差范数是二次相位误差的单峰函数的结论，且在二次相位误差等于零时全变差范数最小。从而利用全变差范数作为性能函数，采用黄金分割最优化算法，通过循环迭代得到自聚焦的SAR图像。实验结果表明此算法在处理二次相位误差时优于相位梯度自聚焦算法。通过算法复杂度的分析表明该算法的计算量约为相位梯度自聚焦的二分之一。

Minimum Total Variation Autofocus Algorithm for SAR Imaging

Minimum Total Variation Autofocus (MTVA) algorithm for SAR imaging is studied. Firstly, the relationship between total variation of SAR image and 2nd-order phase error is analyzed. The results show that total variation is a single peak function of 2nd-order phase error. Total variation reaches minimum when zero phase error. Then total variation norm is used as performance function. Golden selection method is adopted to get focused SAR image. Finally, experimental results show that MTVA algorithm has good performance and is better than Phase Gradient Autofocus (PGA) algorithm. Study of complexity shows that MTVA’s computation amount is nearly half of PGA’s.