Bayesian Classification of Humans and Vehicles Using Micro-Doppler Signals From a Scanning-Beam Radar

摘要

This letter describes a Bayesian formulation for the classification of humans and vehicles using micro-Doppler obtained from a 36 GHz scanning-beam continuous-wave radar. Classification from a scanning-beam system is difficult because of reduced dwell-times and the relatively low amount of time that humans produce strong micro-Doppler signals during typical motion. The classifier analyzes the number of micro-Doppler frequencies present in the return signal over a number of rotations. Experimental results are presented and standard metrics are calculated to evaluate the performance of the classifier. Probabilities of detection near 0.9 are achieved with probabilities of false alarm close to zero.