PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
This paper discusses the various millimeter-wave radar micro-Doppler features of consumer drones and birds which can be fed to a classifier for target discrimination. The proposed feature extraction methods have been developed by considering the micro-Doppler signature characteristics of in-flight targets obtained with a frequency modulated continuous wave (FMCW) radar. Three different drones (DJI Phantom 3 Standard, DJI Inspire 1 and DJI S900) and four birds of different sizes (Northern Hawk Owl, Harris Hawk, Indian Eagle Owl and Tawny Eagle) have been used for the feature extraction and classification. The data for all the targets was obtained with a fixed beam W-band (94 GHz) FMCW radar. The extracted features have been fed to two different classifiers for training (linear discriminant and support vector machine (SVM)). It is shown that the classifiers using these features can clearly distinguish between a drone and a bird with 100% prediction accuracy and are able to differentiate between various sizes of drones with more than 90% accuracy. The results demonstrate that the proposed algorithm is a very suitable candidate as an automatic target recognition technique for a practical FMCW radar based drone detection system.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.