ISAR Image Registration Method Based on PCARANSAC Fusion

Inverse Synthetic Aperture Radar (ISAR) image registration is a key link in 3D imaging and image fusion, but factors such as image sparsity, scattering point migration, and noise increase its difficulty. This paper proposes a registration method combining Principal Component Analysis (PCA) and Random Sample Consensus (RANSAC). The method first performs PCA analysis on strong scattering points to obtain the main direction of the target, achieving rough estimation of rotation angle and translation amount. Subsequently, scattering point matching pairs are obtained based on neighborhood correlation, and the RANSAC algorithm is used to eliminate mismatched points, accurately estimating registration parameters to complete fine matching. Simulation experiments show that this method can still achieve accurate registration under conditions of noise and scattering point loss, with high robustness and stability.