An Effective Parametric Autofocus Algorithm For Earth-based Radar Astronomical Imaging

Earth-based radar astronomical imaging is of great significance to planetary science research. With the improvement of imaging resolution, multiple non-ideal effects appear in the earth-based radar astronomical imaging, such as ephemeris error and atmospheric distortion, which seriously degrades the image quality. Due to the low SNR of echoes, the traditional non-parametric autofocus method is no longer suitable, while the traditional parametric autofocus method is inefficient. Therefore, efficient autofocus techniques are required to obtain well-focused SAR images. In this paper, an effective parametric autofocus technology is proposed. First, the phase error is modeled as a polynomial function of slow-time, which reduces the parameter search dimension. Then, sub-patch and sub-aperture division is performed, multiple sub-patches and sub-apertures are selected to perform back-projection imaging, which reduces the computational complexity of imaging. Finally, the phase error parameters are estimated using the Adam gradient descent method. Moreover, point simulation results and lunar experiment results based on an earth-based radar prototype system validate the proposed algorithm.