Open AccessPCA applied to Data Fusion for Subsurface Target Imaging of Full-polarimetric GPR
Author(s) -
Chaoyang Xue,
Xuan Feng,
Haoqiu Zhou,
Xiaotian Li,
Wei Liang,
Ying Wang
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/660/1/012032
Subject(s) - ground penetrating radar , polarimetry , principal component analysis , remote sensing , sensor fusion , radar , fusion , dimensionality reduction , computer science , polarization (electrochemistry) , artificial intelligence , pattern recognition (psychology) , geology , physics , optics , telecommunications , linguistics , philosophy , chemistry , scattering
Full-polarimetric ground penetrating radar (GPR) can obtain more comprehensive polarization data (called VV, HH, VH) for the same target than traditional commercial radar (only VV). We need to use data fusion technology to combine the polarization information of the three different polarization modes. However, the full-polarimetric GPR data fusion method has one weighted average fusion, which will mask the advantages of full polarization. Principal component analysis (PCA) is a technology of data dimensionality reduction and compression which can use VV, HH and VH as a three-dimensional data to conduct data dimensionality reduction and find the best data fusion results. In order to check the reliability, we obtained the full-polarimetric GPR data of three typical targets in the laboratory for analysis. Then we compare PCA with the weighted average fusion method by using the instantaneous amplitude and conclude that PCA can fuse full-polarimetric GPR data better than weighted average fusion.