Millimeter-wave SAR sea ice classification based on multi-texture feature fusion

Abstract:
This paper focuses on sea ice classification using millimeter-wave synthetic aperture radar (SAR) data to address technical challenges in sea ice monitoring, including sea ice feature diversity and sea ice boundary complexity. The study adopts a new multi-texture feature fusion method based on the attention optimization mechanism, which effectively weights and fuses homologous multi-texture features to optimize sea ice classification performance. The experimental results validate the improvement of the method in terms of sea ice classification accuracy and robustness, and are expected to introduce a new approach in the field of sea ice classification for millimeter-wave synthetic aperture radar.