A NOVEL CROSS-VIEW FUSION APPROACH FOR SHIP TARGETS BASED ON ISAR AND OPTICAL IMAGES

Multi-view inverse synthetic aperture radar (ISAR) imaging and optical imaging have emerged as important techniques for detecting ship targets, which have recently gained attention. Current studies on radar-based target detection and utilization of multi-view information primarily focus on constructing three-dimensional target images from multiple perspectives or treating radar and optical images as distinct domains for conversion within each domain. However, multi-view radar and optical images inherently contain information from different perspectives. During the image conversion process, if the optical image domain includes target perspectives absent in the radar image domain, it becomes essential to address how to transform the radar images' new perspectives into those of the optical domain and achieve fusion between the two domains. To address this problem, this paper proposes a novel transformation method based on the SelectionGAN network that utilizes the unique target perspectives present in optical images to guide the cross-view transformation of radar images. This approach facilitates the fusion of the two image domains, resulting in high-quality fused images that enhance target recognition. Simulations validate the effectiveness of the proposed method.