Ground Penetrating Radar Nondestructive Testing Based on Wavelet Sparse Representation

摘要

Now Ground penetrating radar (GPR) nondestructive testing methods have been applied to many fields of physics. But traditional electromagnetic methods (usually based on least square and local iteration) just roughly give the information of location, level and quality. In this paper we consider inverse electromagnetic problem which is concerned with the estimation of electric conductivity of Maxwell's equations. A wavelet multilevel representation is proposed to inversion of GPR nondestructive testing. Once we decompose the objective functional onto different levels from the smallest to the largest, there are very few local minimum on the largest level component of the problem. Then local convergent Gauss-Newton method could easily find the global minimum on this level which is close to the global optimization solution on the second largest level. So, Gauss-Newton method with initial value which is solved on the largest level has serious possibility to find the global minimum of the second largest level. Repeating this step one could find the global optimization solution of the original inverse problem. On each level, the stable and fast local convergent Gauss-Newton method is carried out. Results exhibits clear advantages over damping Gauss-Newton method and testify that it is an available method, especially on aspects of wide convergence and precision.