Ship Recognition for Complex SAR Images via Dual-Branch Transformer Fusion Network

Ship recognition in synthetic aperture radar (SAR) is an essential challenge in SAR image interpretation. The measured SAR ship targets often contain complex backgrounds such as port facilities and neighboring ships, which are easy to interfere with the model and affect the recognition performance. To address this issue, a SAR ship recognition method with a complex background based on a dual-branch transformer fusion network is proposed in this letter. First of all, a dual-branch feature extraction and fusion architecture is designed in this letter, including significant feature extraction (SFE), global feature extraction (GFE), and dual-branch feature fusion (D-BFF). Specifically, the SFE effectively extracts the most discriminative local fine-grained features of the ship target using multilayer convolution of significant regions. The GFE captures global semantic information by residual module optimization. In addition, combined with the self-attention in the transformer block based on cross-attention and position encoding, the effective fusion of SFE and GFE is realized in D-BFF. Finally, extensive experiments are carried out based on the Gaofen-3 seven-category dataset (anyone can get the dataset after sending the application e-mail). The results reveal that the proposed method can achieve a recognition accuracy of 75.55%, which is significantly superior to other algorithms.