Automated shape classification of buried objects with low frequency multistatic sonar

In this paper a novel and simple technique is presented that exploits multi-platform sonar capabilities for automated target shape classification. This technique is based on the relation between the target beampattern and the size and shape of the object. The performance are evaluated in terms of mean square error of the size estimator and probability of correct classification. This technique is shown to offer improved performance and lower computational load and physical constraints compared to the imaging-based classification techniques.