AUTOMATIC TARGET CLASSIFICATION USING HF MULTIFREQUENCY RADARS.

摘要

The classification of radar targets such as aircraft and ships using lower resonance-region radar returns has been of significant interest in recent years. The H.F. band is in the resonance region of such targets. The probability of misclassification depends upon the post-processing signal-to-noise power ratio.; Current techniques for measuring and processing the amplitude and phase of H.F. radar returns are reviewed. The post-processing SNR is determined as a function of coherent observation time. Based on features extracted from the radar returns, a frequency-domain Nearest-Neighbor classification algorithm and a time-domain Cross-Correlation classification algorithm are designed. A set of radar backscatter measurements of scale model ships and aircraft is used to generate a data base for testing the classification algorithms. Statistical techniques are applied to determine the probability of misclassification as a function of the post-processing SNR. The probability of misclassification for a reasonable amount of coherent observation time is determined.