Clutter Suppression for Lunar Penetrating Radar via Blind Sparse Deconvolution with Spatial Regularization

The Chang'E-4 mission marks a milestone in space exploration since it was the first time that a human-made object ever soft landed on the far side of the Moon. The rover of the mission, Yutu-2, carries two Ground-Penetrating Radar (GPR) systems designed to map the lunar subsurface and determine its dielectric properties. To that extent, various processing techniques have been suggested to enhance interpretation and increase the overall quality of the data. This paper introduces a method for deriving the 2D subsurface reflectivity series using blind sparse deconvolution with spatial regularization. Blind sparse deconvolution increases the resolution of the radagram, and provides a comparative assessment of the dielectric properties of different layers. The derived reflectivity series and source wavelet are then used to reconstruct the B-Scan, excluding any sources of noise and clutter that cannot be captured by the sparse model. This effectively suppresses clutter without compromising the amplitude and resolution of targets' reflections. The proposed scheme is tested on both real and numerical data, highlighting its effectiveness in enhancing the interpretation of the Chang'E-4 data, and demonstrating its broader value for planetary radar in general.