Algorithm on the Estimation of Residual Motion Errors in Airborne SAR Images

Abstract

Due to the lack of accuracy in the navigation system, deviations in the order of centimeters between the real trajectory and the measured one, called residual motion errors (RMEs), frequently appear in synthetic aperture radar (SAR) images. They usually cause azimuth defocusing and phase errors. For very high-resolution SAR imaging and repeat-pass SAR interferometry, the RME must be estimated and compensated. In the literature, the algorithms used to estimate such errors are mainly through interferometry. The error difference for the SAR image pair is calculated, and then the correction is applied either on one image, or alternatively fractions of it on both images. However, the accuracy of this approach is greatly deteriorated by the decorrelation and ground movements. In this paper, we associate the phase difference between two adjacent subaperture images with the second derivative of the RME, and propose a new algorithm to estimate the RME for individual SAR image. It exploits point-like targets distributed along the azimuth direction, and not only corrects the phase error, but also improves the azimuth focusing. Therefore, besides in airborne repeat-pass interferometry, the algorithm can also be applied in very high-resolution SAR imaging. Simulated and real SAR data are used to demonstrate its feasibility and accuracy. Its limitations and extensions are also discussed at the end of the paper.