SAR Automatic Target Recognition with Multiview Target Information Fusion and Ensemble Learning

Although synthetic aperture radar (SAR) could see through night, clouds and smoke, SAR automatic target recognition still faces many daunting technical challenges, such as the huge cost in data acquisition and the potential risk of falling victim to the electronic attacks exerted by the adversaries. Bistatic SAR can obtain richer information from multiple angles, and offers better anti-jamming and anti-interception performance than its monostatic counterpart. To illustrate the potential advantage that could be brought by combining the target information acquired from multiple elevation angles, we leverage the proprietary NUAA MiniSAR ground military vehicle dataset and propose a novel multiview target information fusion scheme based on decision-level ensemble learning. The effectiveness of the proposed method is verified by a series of ingeniously-devised experiments, which lays the foundation for future research on monostatic-bistatic SAR target feature fusion.