Filtered Convolution for Synthetic Aperture Radar Images Ship Detection

Synthetic aperture radar (SAR) image ship detection is currently a research hotspot in the field of national defense science and technology. However, SAR images contain large amount coherent speckle noise, which poses significant challenges task detection. To address this issue, we propose filter convolution, novel design that replaces traditional convolution layer suppresses noise while extracting features. Specifically, kernel comes from input generated by two modules: kernel-generation module local weight generation module. The dynamic structure generates kernels using or feature information. based on statistical characteristics features used to generate weights. introduction weights allows extracted more characteristic information, conducive images. In addition, proved fusion proposed can suppress image. experimental results show excellent performance our method large-scale dataset-v1.0 (LS-SSDD-v1.0). It also achieved state-of-the-art high-resolution dataset (HRSID), confirmed its applicability.