Classification of Radar Targets via Distribution Matching of Late-Time Resonance Parameters

A promising non-imagining approach to the classification of radar targets is to use the frequencies and attenuation rates of the resonant modes that present during a target’s late-time response (LTR) as features. Unfortunately, the estimation of these resonance parameters is rather sensitive to noise. However, we observe that when a large number of measurements of the LTR can be taken in a short time, the probability distribution of the estimates of the parameters can be estimated, and then matched against a database of such distributions. That has the potential to reduce the sensitivity of the classification problem to noise. In this paper, we develop a pragmatic approach to target classification using this distribution-matching approach, and demonstrate its effectiveness through physical experiments. The proposed approach is shown to be highly robust to environmental clutter and somewhat robust to target orientation.