Multi-polarization Data Fusion Analysis of Full-Polarimetric Ground Penetrating Radar

Abstract

Full-Polarimetric Ground Penetrating Radar (FP-GPR), compared to traditional single-polarimetric GPR, can obtain more comprehensive polarization data (such as VV, HH, and VH) for the same target. To ensure a more comprehensive targets’ image identification, data fusion technology is applied to FP-GPR so as to combine the polarization information of three different polarization modes. However, weighted average fusion is usually employed in FP-GPR data fusion, since it masks the advantages of full polarization and is unable to simultaneously adapt to different target scattering mechanisms. Based on Principal Component Analysis (PCA), Laplacian Pyramid (LP), and multi-scale Wavelet Transform (WT), this research proposes three FP-GPR data fusion methods. To check the reliability of several data fusion methods, we obtained FP-GPR data representing three different target scattering mechanisms in the laboratory and, then, compared the weighted average fusion method with the other three methods using instantaneous amplitude and gradient. The result shows that the three methods were better than the weighted average fusion and that they can be adapted to different target scattering mechanisms. However, PCA was used to fuse the unknown target scattering mechanisms. Finally, PCA fusion is applied to actual ice fracture data imaging, as it produces a better fusion effect than that of weighted average fusion.

Keywords

• full-polarimetric ground penetrating radar (fp-gpr)
• data fusion
• principal component analysis (pca)
• laplacian pyramid (lp)
• wavelet transform (wt)
• target imaging