Compressive Sensing Application to Real-Beam Inverse Synthetic Aperture Radar (ISAR) Imaging and Automatic Target Recognition

During the final stage of flight, just before impact, the radar-equipped missile that was flying at a squint angle to the target to obtain ISAR/SAR imagery must turn the missile velocity vector directly at the target, which eliminates high-quality ISAR images and the radar is forced into a real-beam degraded imagery mode. Our proposed approach exploits the application of CS signal processing to extract the ship features and dominant scatterers from the limited radar pulses imposed by real-beam mode. We propose to develop a novel compressive analog-to-feature converter that extracts the salient features and dominant scatterers from the limited radar pulses directly from the analog signal within the CS framework. The extracted scatterers range-Doppler locations are then tracked over multiple frames of radar pulses as the weapon flies in to assure consistency and provided to a maximum a posteriori (MAP) estimator that then estimates the scatterers 3-D locations (i.e., point cloud image) by solving for the unknown coefficients in the dynamical equations of motion of the ship. We then drape the extracted features over the point cloud image using standard photogrammetry approaches and apply an iterative closest point approach to classify the ship by matching against existing CAD models.