Performance evaluation of radar NCTR using the target aspect and signature

In this paper, we present a framework for assessing and comparing the performance of non-cooperative target recognition methods that are based on radar signatures. The proposed framework aspires to quantify the performance considering the dynamic nature of the operating conditions. When evaluating the recognition performance, the training data is generated according to the operational scenario which specifies for instance the values for the true aspect angle, at which the target is observed, and the ability to estimate it, e.g. by a radar tracker. The performance of different classifiers can be assessed and compared considering the varying ability to observe the location and the orientation of the target. The proposed framework is demonstrated with the radar cross section as the feature for classification using values on a single frequency and high range resolution profiles.