Maximum likelihood autofocusing of radar images

In ISAR and SAR imaging, the relative motion between the radar and the target must be known precisely otherwise the synthetic aperture becomes defocused, producing a radar image with severe cross-range blurring. The paper estimates changes in a target's range using maximum likelihood estimation. A two-stage algorithm to find the ML estimator is proposed which uses the chirp-Z transform for coarse estimates and an iterative phase estimator for fine estimates. The effectiveness of the ML-based approach is demonstrated in eliminating motion blur from a simulated ISAR image. Finally, various motion estimation schemes proposed in the ISAR literature are shown to be equivalent to partial implementations of the ML estimator.