The alignment problem for 3D ISAR imaging with real data

The alignment between the point-like 3D inverse synthetic aperture radar (ISAR) reconstruction produced by 3D ISAR imaging and model of the target is an essential issue in target recognition by 3D ISAR reconstruction, which contains coarse alignment and accurate alignment. The coarse alignment can be accomplished by principal component analysis (PCA). However, PCA may suffer from the 180-degree ambiguity problem due to the uncertainty of the orientations of principal components. In this paper, an effective and robust approach making use of the orthogonality among the principal components and k-d tree is proposed to address the ambiguity problem. The experimental results of the real measured data verify the validity of the proposed method.