In high-resolution inverse synthetic aperture radar (ISAR) imaging, the rotational motion of the targets tends to introduce the time-variant Doppler modulation in the echo, which acts as the range-variant phase errors in the phase history. Moreover, the performance of translational phase error correction may be dramatically degraded without properly considering the range-variant phase errors. In this study, a joint approach of translational and rotational phase error corrections is introduced into high-resolution ISAR imaging. In the procedure, the joint phase error correction is modelled as that of range-invariant and range-variant phase errors using a metric of minimum entropy. Then, the minimum-entropy optimisation is solved by employing a coordinate descend method based on quasi-Newton solver. In comparison of the conventional methods, the proposed approach in this study promises a better performance of phase error correction with a higher efficiency. Finally, experiments based on simulated and measured data are performed to confirm the effectiveness of the proposed algorithm.