All-Aspect Maritime Automatic Target Recognition

The Navy desires a method to exploit radar reflections from shipboard antennas to classify vessels regardless of their aspect angles, in particular near broadside where other ISAR-based classification methods perform poorly.  EMSI proposes to demonstrate the feasibility of extracting features from both rotating antennas and non-rotating electronically scanned arrays (ESAs), and of using those features for classification.Scatterers on rotating antennas generate micro-Doppler returns spread widely in crossrange and over the range extent of the antenna.  Micro-Doppler information is commonly extracted from a spectrogram created by applying a Short Time Fourier Transform (STFT) to the range interval containing the antenna.  However, the STFT presents a trade-off between time and frequency resolution.   Bilinear time-frequency transforms improve the time-frequency resolution but suffer from nonphysical cross-terms.  We propose an alternative approach that compresses all of the responses simultaneously, as well as an approach to recognize returns from a stationary ESA. The proposed effort will use simulated and collected ISAR data to establish feature extraction accuracies and uncertainties, will create a classification database of representative combatants and commercial vessels, and will estimate the achievable level of classification.