Discrete Tomography Approach for Subsurface Object Detection by Artificial Neural Network

The problem of recognition of the depth of hidden object by ultrawideband electromagnetic field irradiation, receiving of the reflected wave, processing of the signal on the basis of tomography approach, and its analysis by artificial neural network is considered. The irradiated field is a transient plane wave. The hidden underground object is a metal tube. The electromagnetic problem is solved by FDTD method. The reflected field is received by set of probes under the ground surface. Tomography approach consists in the forming of new data set for artificial neural network by increasing of the signal using geometrical peculiarities of the electromagnetic problem. The influence of the position of time window for the recognition result is studied.