An Efficient Scaled Maximum Likelihood Algorithm for Translational Motion Estimation in ISAR Imaging

In ISAR imaging, the relative motion between the target and the radar must be known precisely to produce focused radar images. The translational motion of the target must be compensated for to use only the rotational motion around a fixed centre point for the imaging of the target. An efficient implementation of a maximum likelihood (ML) algorithm for translational motion estimation based on the Chirp-Z transform is described. If the line of sight vector from the radar to the target is not within the rotational plane of the object, or the rotational plane changes during the observation time, and strong reflectors tend to bias the estimate of translational motion. A scaling of range profiles is shown to reduce the bias. The shear average algorithm is similar to the algorithm described here, but it only estimates translational motion to within half the carrier wavelength. Simulated and experimental data are used to show the effectiveness of the algorithm. An image sharpness measure is used to indicate the effects of scaling as a preprocessing step on experimental data. All results are compared to those obtained by shear average and prominent point processing techniques.