Improvements of radar clutter classification in air traffic control environment

The use of adaptive technologies may prove useful in the processing of radar signals. The proposed radar clutter classificator is aimed to improve the detection of snow clutter presence in data acquired by a ground radar system in an air traffic control environment. Each plot detected in the radar image is processed in order to extract a series of features which are then used to discriminate between targets or snow clutter. The data set used in simulations has been extracted from a measurement campaign carried out in presence of snow in Italian airports, with the help on an expert operator who manually classified a set of plots as target, or clutter. Two sets of features have been tested, one of which is derived from moments of inertia, and the second one is a specifically designed feature vector, which has been selected as to follow the reasoning scheme of a trained human observer. Three different classifiers have been tested and compared in the final stage: a Bayes classifier, a multilayer perceptron and a radial basis function network. Results indicate the best configuration exhibits a correct classification rate of about 95%.