| This thesis addresses the problem of how to remove the smear in a high-resolution airborne spotlight SAR image, which is caused by the target's constant translational motion. Applications include automatic ship identification from real-time airborne spotlight SAR images. A mathematical model for the two-dimensional impulse response of the image smearing process, which is a function of the target velocity is developed. An algorithm employing a Kalman filter is used to estimate the target velocity. With the estimated target velocity, the smear is removed by inverse filtering the phase of the smeared image. The main part of this thesis proposes a new algorithm of target velocity estimation, which uses target centroid measurements taken from subaperture images. Radar imagery simulations are used to study the relationship between target centroid measurement noise and target velocity. No a priori knowledge of target velocity is required for the velocity estimation in this algorithm. To evaluate the performance of the estimator, mismatched measurement noise is used in the Kalman filter, as we do not know the true target velocity. The target velocity variance is calculated analytically for comparison purposes. The accuracy of velocity estimation is found to be inversely proportional to the magnitude of the true velocity. It is shown that with appropriate subaperture integration time intervals, it is possible to restore the smeared images despite the mismatched measurement noise. |
