A Frequency-Phase Gradient Autofocus Algorithm for Excessive Migration in UAV SAR Images

Synthetic Aperture Radar (SAR) installed on unmanned aerial vehicle (UAV) is a high-efficiency but low-cost remote sensing system. On the one hand, UAV is very sensitive to atmospheric turbulences, which causes serious trajectory deviations and makes high resolution imaging a challenging task. On the other hand, limited by the payloads, UAV may not carry high-accuracy inertial navigation systems (INS)/global positioning systems (GPS). Especially for Very High Resolution (VHR) condition, residual range migration can exceed range resolution, in which case conventional autofocus algorithms such as phase gradient autofocus (PGA) fail. In this paper, applying frequency-phase gradient estimation and corresponding correction based on building signal error model, a new combinatorial method, which can be regard as the extending PGA, has been developed. Here, we estimate range misalignment by fast-time split-band signal difference at each slow time, and obtain phase error through calculating actual range cell phase shift gradient at each azimuth time. The proposed algorithm is validated by practical experiments.