Subsurface cylindrical object location and material inversion from GPR data-based online SVR

In this paper, an integrated approach is proposed to inverse geometrical and electrical properties of cylindrical objects buried in isotropic and homogeneous medium using Ground penetrating radar imagery. As the peak amplitude of receiving signals depend on electrical properties of objects and background, objects depth, there exist a nonlinear relationship between them. The integrated approach is composed of four steps. First, a symmetry-based algorithm is used to obtain objects depth information. Secondly, different velocity models and the root-mean-square error are applied to perform the estimation of relative permittivity of background. Thirdly, Hough transform is used to estimate the size of objects. After these, the multivariable recursive problem can be reduced to a single-variable parameter inversion task. Finally, online support vector regression is adopted to inverse the object material. The results based on simulated data demonstrate that the proposed approach can efficiently and precisely solve these problems.