Enhanced CRS-Based Wavefield Separation and WNE-Based Velocity Estimation for UAV-Borne GPR Data Processing

The application of unmanned aerial vehicle (UAV)-borne ground penetrating radar (GPR) for subsurface detection has garnered considerable attention. The estimation of subsurface velocity using diffraction-focusing-based methods is integral to the reconstruction of subsurface structures. Nevertheless, such methods are challenged by the influence of reflections on diffraction analysis, and regions of noninterest impede the accuracy of media velocity estimation. In this study, we introduce a processing scheme to solve these problems. First, a common reflection surface (CRS)-based approach is implemented to identify and mitigate reflections from raw GPR B-scan data, leveraging a directional filter function derived from CRS wavefront characteristics. The optimization process is refined by selecting key parameter ranges based on the first Fresnel zone. Second, a weighted negative entropy-based velocity estimation approach is realized by integrating a weighting function into the statistical negative entropy (NE) calculation. These weight factors, derived from the air-coupled GPR travel time model, accentuate hyperbolic features, thereby elevating resolution in time-velocity (t–v) panels. The efficacy of our scheme in separating wavefields is demonstrated by the diffraction-to-reflection ratio (DRR) of 6.544 in simulations and 4.510 in field experiments, representing improvements of approximately fivefold and threefold, respectively. The resolution of t–v panels with the proposed scheme increases by more than 20%, highlighting its superiority in estimating media velocity. In the reconstructed images, the maximum depth-estimation error is 5.22%, and the root-mean-square relative error (RMSRE) is 3.37% in simulation. In the field test, the signal-to-noise ratio improvement shows the target’s return power at 14.8 dB above the noise floor, and the depth-estimation error is 4.41%.