An Effective Back-Projection Autofocus Algorithm for Earth-Based Radar Lunar Imaging

The back-projection (BP) algorithm has been regarded as a robust high-resolution imaging algorithm, particularly suitable for Earth-based radar lunar imaging. To offset phase errors induced by non-ideal effects, autofocus is essential to obtain well-focused lunar images. However, traditional back-projection autofocus algorithms are often computationally burdensome under long synthetic aperture time. In this paper, an effective back-projection autofocus algorithm is developed for Earth-based radar lunar imaging. To ensure that the defocusing caused by phase error lies in the azimuth direction of the BP image, the delay-Doppler coordinate system is introduced into the BP algorithm, and lunar images are generated on the delay-Doppler grid arranged on the lunar surface. Then, the Fourier transform relationship between the BP image and its wavenumber spectrum is established to facilitate the application of phase gradient autofocus (PGA) algorithm to the BP image. To further improve the estimation accuracy of phase error, the spectral characteristics are analyzed in detail. A multi-subband autofocus method based on local scene is proposed to mitigate frequency-dependence of phase error and space-variance of wavenumber spectrum. The processed results of real lunar data validate the effectiveness and efficiency of the proposed algorithm.