Phase-Congruency Reflector Detection of Orbital Radar Data via Shearlet Transform on Shannon-Gabor Wavelet

The investigation of subsurface stratigraphy is crucial for understanding the evolution of planetary bodies. Successful missions have underscored the indispensable role of orbital radar sounders in exploring the Martian subsurface. The abundance of sounding radar data has spurred the development of automatic techniques for identifying subsurface reflectors, to reduce the reliance on manual picking. However, the prevailing detection algorithms face significant challenges in addressing the variability of radar data across different geological contexts or from various instruments, especially with strong noises. These obstacles complicate the accurate extraction of subsurface layer structures and echo power, thereby impeding our understanding of geological evolution. Here, we introduce a novel approach for automatically detecting subsurface reflectors from orbital radar data, leveraging a combination of phase congruency and the shearlet system for multiscale analysis. It employs the Shannon-Gabor (SG) wavelet as the principal component in its detection framework, which proves to be more effective than the traditionally used Mexican hat wavelet, as evidenced by synthetic data experiments. Tests on both shallow radar (SHARAD) and mars advanced radar for subsurface and ionosphere sounding (MARSIS) radargrams have shown that our method is able to detect weak, quasi-scattering, and highly undulating reflectors, while maintaining a desirable level of precision comparable to that of human analysis. The proposed method demonstrates versatility across various radar datasets and imaging scenes, showing great promise for revealing additional information about planetary subsurface structures and supporting data analysis for future spaceborne missions.