Wide-Angle (WA) synthetic aperture radar (SAR) has shown remarkable performance in high-resolution mapping, and its mostly used imaging algorithm, fast factorized back projection (FFBP), is efficient, robust, and of low computational complexity. However, trajectory deviations and the system calibration error, introduced by low measurement accuracy, dramatically degrade FFBP's performance. This article proposes the modified factorized geometrical autofocus (MFGA) method for WA-SAR to address the above problems in FFBP. MFGA implements the phase gradient algorithm on defocus subimages at first. Then, a factorized geometrical error hypothesis between subimages is proposed. And the corresponding defocus factors are classified as three independent parts: image distortion, spectrum migration, and phase error. To deal with those problems, MFGA introduces image registration techniques and a maximum sharpness method to calibrate image distortion and phase error. Moreover, in MFGA, based on minimum entropy and least square method, a Doppler spectrum migration correction algorithm is proposed to correct spectrum migration. In the FFBP chain, MFGA is used on subimages refocus and fusion until obtaining a full-resolution image. In our experiments, we compared MFGA and other time-domain autofocus algorithms using simulated data and real data obtained by helicopter and airship with false trajectory and system calibration parameters. The results show that MFGA performs better in terms of the peak to side-lobe ratio, the azimuth resolution, the refocused images' entropy, and processing time consumption. The better performance demonstrates MFGA's advantages in addressing trajectory deviations and the system calibration error for WA-SAR.

中文翻译：

---

一种改进的广角合成孔径雷达分解几何自动对焦方法

广角（WA）合成孔径雷达（SAR）在高分辨率映射方面表现出卓越的性能，其主要使用的成像算法快速分解反投影（FFBP）高效、稳健且计算复杂度低。然而，由低测量精度引入的轨迹偏差和系统校准误差极大地降低了 FFBP 的性能。本文提出了改进的因子分解几何自动对焦（MFGA）方法用于 WA-SAR，以解决 FFBP 中的上述问题。MFGA首先在散焦子图像上实现相位梯度算法。然后，提出了子图像之间的分解几何误差假设。并将相应的散焦因子分类为三个独立的部分：图像失真、光谱偏移和相位误差。为了解决这些问题，MFGA 引入了图像配准技术和最大锐度方法来校准图像失真和相位误差。此外，在MFGA中，基于最小熵和最小二乘法，提出了一种多普勒频谱偏移校正算法来校正频谱偏移。在 FFBP 链中，MFGA 用于子图像重新聚焦和融合，直到获得全分辨率图像。在我们的实验中，我们使用模拟数据和直升机和飞艇获得的具有虚假轨迹和系统校准参数的真实数据来比较MFGA和其他时域自动对焦算法。结果表明，MFGA 在峰旁瓣比、方位角分辨率、重新聚焦图像的熵和处理时间消耗方面表现更好。更好的性能证明了 MFGA'