Neural networks for the detection of buried plant

Systems based upon radar techniques have been used over the past twenty years or so as a means of obtaining information without the need for excavation. The areas of application include geological prospecting, archaeological investigations and the detection of buried pipes and cables (plant). The paper deals with the use of an artificial neural network for the detection of buried plant. The problem is split into two distinct parts, a pre processing stage which essentially extracts features in the form of spectral coefficients followed by the neural network classifier. The latter successfully discriminates between surface and target segments present within the radar traces captured by a ground probing radar. The neural network has numerous advantages over the conventional method of envelope filter detection usually adopted and these are demonstrated by its performance on data from three test sites