Physics-Based Maritime Target Classification and False Alarms Mitigation

RDRTec proposes to study the feasibility of an innovative maritime automatic target recognition (ATR) system concept called Physics Inspired ISAR ATR System (PI2AS) that enables large scale target recognition on low-power mobile computing platforms while dramatically improving probability of correct classification (Pcc) and probability of false classification (Pfc). We have performed an initial demonstration of our algorithm against inverse synthetic aperture radar (ISAR) images of maritime targets and showed over two orders of magnitude improvement in processing requirements and reduction in Pfc while maintaining comparable Pcc to state-of-the-art ATR algorithms. Our integrated approach will enable fast generation of target signatures for rapid training of new targets while exploiting the physics of radar scattering within a Bayesian compressive manifold learning ATR approach to dramatically improve rejection of non-targets (i.e., false targets and decoys) and classification of high-valued targets of interest in real-time on size weight and power (SWaP) constrained platforms.