Abstract: Due to the presence of multiple targets, the coherence of radar echoes is disrupted, rendering traditional inverse synthetic aperture radar (ISAR) imaging methods ineffective. This paper addresses the challenges of spectrum aliasing and separation difficulties in ISAR imaging of formation targets by analyzing their formation and motion characteristics. We propose a high-density multi-target ISAR imaging method based on Keystone transform and high-order rotation phase compensation. First, the echoes of multiple targets are processed via blind signal separation to generate an initial composite image of the target group. The Keystone transform is applied to migration through resolution cell (MTRC) and minor radial velocity variations among individual targets. Next, the equivalent rotation angle and radial position of the target group′s center are estimated using Brent method, and a high-order Doppler phase factor is constructed to compensate for Doppler aliasing caused by long observation times. Finally, fine-resolution imaging of each target is achieved through DBSCAN clustering. Simulation results demonstrate that the proposed method enables well-focused imaging of group targets while accurately resolving their relative spatial relationships. Furthermore, auto-focused single-target images deliver significantly finer details, thereby validating the correctness and effectiveness of the proposed approach.