FEF-Net: A Deep Learning Approach to Multiview SAR Image Target Recognition

Synthetic aperture radar (SAR) is an advanced microwave imaging system of great importance. The recognition real-world targets from SAR images, i.e., automatic target (ATR), attractive but challenging issue. majority existing ATR methods are designed for single-view images. However, multiview images contain more abundant classification information than which benefits and recognition. This paper proposes end-to-end deep feature extraction fusion network (FEF-Net) that can effectively exploit boost the performance. proposed FEF-Net based on a multiple-input structure with some distinct useful learning modules, such as deformable convolution squeeze-and-excitation (SE). Multiview be extracted fused these modules. Therefore, excellent performance achieved by FEF-Net. superiority was validated experiments moving stationary acquisition (MSTAR) dataset.