A statistical radar clutter classifier

Abstract

An algorithm that successfully classifies radar clutter into one of several major categories, including bird, weather, and target classes, is described. Parametric Bayes classification is applied to a set of features derived from the reflection coefficients computed using the multisegment version of Burg's formula. These coefficients are then transformed and grouped to meet the requirements for multivariate Gaussian behavior. The addition of two amplitude-related features aids in distinguishing between point targets and distributed clutter. Average probabilities of correct classification of 70% to 90% have been found when testing the classifier on recorded radar data.<>