Abstract: Inverse synthetic aperture radar (ISAR) imaging of the non-cooperative maneuvering multi-target is a challenging task due to its time-varying orientation and the relative motion among targets. Conventional motion compensation methods assume that targets' positions are almost fixed during the imaging time. In this paper, the model of relative motion targets is established and the effect of relative motion for ISAR image is analyzed. A new algorithm for relative motion compensation based on Keystone transformation and Autoclean algorithm is proposed. The imaging results using simulation and real data validate the new relative motion compensation method.