A Doppler-based automatic target classifier for a battlefield surveillance radar

This paper describes the automatic target classifier which has been introduced into the Man Portable Surveillance and Tracking Radar (MSTAR) battlefield surveillance as part of its mid-life improvement. It discusses the background to the use of Doppler classifiers for radars of this type, based on the already-prevalent use of audio as a classification aid. It then discusses the principles of the classifier used in the MSTAR radar. This uses the Doppler spectrum to form a feature vector and uses a hierarchical Fisher linear discriminant to resolve the different classes. The training and testing of the classifier are discussed, together with the performance achieved.