Subsurface Object Detection in Randomly Inhomogeneous Medium Model

The problem of hidden underground object detection in random inhomogeneous medium with the help of ground penetrating radar and artificial neural network is solving in the work. The irradiator generates a short Gaussian impulse electromagnetic wave which penetrates into the ground surface and propagates through the random inhomogeneous ground modelled by homogeneous medium with random inclusions with random distribution of placement, size and permittivity. The field reflected from buried object, model of antipersonnel mine, is detected by the system of four receiving antennas. The electromagnetic problem is numerically solved by FDTD method. Four-element antenna system is utilized to compensate the reflections from receiving antennas and the ground surface. Discretized result signal is used as input signal for artificial neural network. The artificial neural network is learnt to recognize hidden objects in homogeneous medium. The network is applied for classification of an object in heterogeneous medium and its coordinates in presence of Gaussian noise in received signals.