Bird body and wing-beat radar Doppler signature separation using sparse optimization

Radar bird detection and discrimination has many civilian and non-civilian applications such as collision avoidance, false alarm reduction for detection radars, stealthy target detection, classification of military unmanned aerial vehicles (UAVs) and civilian drones, and conservation ecology. In order to develop new and improve existing detection and discrimination algorithms, this paper proposes a feature extraction technique in which the wing-beat Doppler radar signature of a bird is separated from its respective body signature. More specifically, we propose non-linear morphological component analysis (MCA) using invertible short-time Fourier transform (STFT) for feature extraction. The method is applied to the Peregrine falcon data measured by Alabaster et al. (2012) resulting in successful separation of the aforementioned signatures.