Noise robust classification of moving vehicles via micro-Doppler signatures

For robust classification of moving wheeled and tracked vehicles using returned micro-Doppler signals within short dwell time, the influence of receiver white noise and low spectrum resolution are encountered. In this paper, noise reduction and super-resolution are realized simultaneously via a redundant dictionary based l₁ -norm optimization method. Experiments based on the measured data are presented, including the analysis of noise reduction performance, and the evaluation of classification robustness for different signal-to-noise ratio cases. The experimental results are also compared with related methods.