Track-Feature-Based Target Classification in Passive Radar for Low-Altitude Airspace Surveillance

The low-altitude airspace is becoming more and more busy. To ensure low-altitude safety, non-cooperative systems (e.g. the passive radars) should be developed in addition to the cooperative ones. Nevertheless, several classes of targets (e.g. aircrafts, birds, and vehicles) may be simultaneously sensed by non-cooperative systems. They should be reliably classified before some pertinent countermeasures are taken. In this paper, we propose a track-feature-based target classification method in passive radar. First, the kinematic features, bistatic radar cross section (BRCS)-related features, and track morphology features are constructed from original track information. The significance of the above handcrafted features are then evaluated via neighborhood component analysis (NCA)-based algorithm to select the most discriminative ones for further use. Then the statistical knowledge (SK)-based classifier and back propagation neural network (BPNN)-based classifier are successively developed to realize target classification. The field experimental results demonstrate the effectiveness and practicality of the proposed method.