MSF-IOF: A Novel ISAR-and-Optical Image Fusion Method Based on Features of Multisubbands

Inverse synthetic aperture radar (ISAR) images and optical images exhibit a certain degree of complementarity due to their imaging in different microwave frequency bands. To enhance the representation capability of spacecraft structures and details, we propose a novel ISAR-and-optical image fusion method based on the features of multisubbands. First, given the different imaging planes of spacecraft in ISAR and optical images, we propose an ISAR-and-optical image registration method that combines keypoint detection and homography transformation, based on the obvious geometric features of the spacecraft. Second, the multiscale decomposition of the source images is achieved by the nonsubsampled shearlet transform (NSST). Subsequently, we propose a novel activity level measurement function based on brightness, contours, and textures to achieve the fusion of low-pass subbands. Simultaneously, the texture of high-pass subbands is effectively fused based on the parameter-adaptive dual-channel pulse-coupled neural network (PADCPCNN). Finally, the fused image is obtained by the inverse-NSST. Compared to the existing state-of-the-art fusion methods, the proposed method has a better performance in both qualitative and quantitative evaluations across multiple imaging instants for different satellite models.