Detection of buried dielectric anomalies by means of the bispectrum method

The development of a target decision system capable of detecting various types of buried dielectric anomalies, such as land mines, under different environmental conditions is addressed. The authors develop a detection and classification scheme for different buried dielectric anomalies. The 2-D extension of a bispectrum method is employed. The feature vectors obtained using this method are used to train and test a three-layer neural network detector. Target detection is based on the shape of the backscattered signals from the anomalies. Detection/classification is performed using a multilayer perceptron neural network. Results of simulations on two types of target are presented