A ground penetrating radar imaging method based on probabilistic grid

The traditional ground-penetrating radar (GPR) imaging methods face challenges such as artifact interference, obscured target signals due to background clutter, and difficulties in detecting weak-reflective targets. To address these issues, this paper proposes a probabilistic grid-based imaging (PGI) method for GPR. This approach discretizes the detection area into 2D grid units, acquires target feature points through phase-based angle estimation and two-way travel time analysis from GPR profile data, and maps them to corresponding grids. Subsequently, by utilizing Bayesian theory to update network weights, we construct a probability distribution model that characterizes underground structures, thereby obtaining PGI imaging results. Both simulation experiments and real road tests demonstrate that compared with traditional methods, the proposed PGI method can effectively suppress artifacts, distinguish subsurface targets, enhance abilityanti-interference ability, achieve higher imaging resolution and facilitate radar profile interpretation later.