3D ISAR imaging: Multi-view image fusion problem

Inverse synthetic aperture radar (ISAR) is capable of generating the three-dimensional (3D) reconstructions of the non-cooperative target via a dual interferometric system, which is called 3D InISAR imaging. The output of 3D InISAR is a 3D point-like image of the target, namely point cloud, each point contains the 3D coordinates (range, cross-range and height with respect to the image plane). Compared to the traditional two-dimensional ISAR imaging, 3D InISAR can avoid the unknow image projection plane and cross-range scaling issues. However, due to some limitations of 3D InISAR imaging, such as scatterer scintillation, self-occlusion and so on, some information of the target in 3D ISAR reconstruction from a single observation view may be missing. In order to obtain a more complete 3D ISAR reconstruction, an incoherent multi-view image fusion method based on the principal component analysis and iterative closest points algorithm is proposed. Results of the simulated data verify the effectiveness of the proposed method.