This dissertation presents overall research results about the performance improvement method of AF and BPA with real SAR images under highly squinted mode operation. BPA is the robust algorithm in the existence of large motion and highly squinted angles. However, it always suffers from very high computational burden. FBPA is an alternative approach to reduce the computational complexity, but its performance could be degraded under non-linear flight trajectory.
Firstly, we proposed the motion compensation method to use target locations on ground plane in case of FBPA. To apply the proposed method to FBPA, SAR image directly formed on the ground plane without degradations due to imperfect motion compensation.
Secondly, Autofocus(AF) is necessity to compensate the SAR performance degradation by motion measurement residual errors in the SAR imagery. When BPA is used to form SAR image on the ground acquired under the highly squinted operation, preprocessing method of AF for BPA is studied. image backplane rotation and doppler alignment method considering squint angle are proposed.
Thirdly, Several famous AF methods are fully reviewed and investigated about the robustness of them when those are applied to the real SAR raw data. We explain the drawbacks of conventional AF methods and propose the revised methods to show the robust results in any real SAR acquisition cases.
Finally, spatially variant phase error(SVPE) is considered to handle the high resolution wide swath situation when the swath width is wide and motion measurement error is big. The AF approach based on partitioned SAR images is suggested to compensate SVPE effectively. In addition, the cross correlation technique is proposed to estimate the azimuth shift offset for mitigating the image mismatching situation.
All of proposed methods are verified by using real SAR data acquired in a flight test and demonstrate superior performance improvement in various acquisition scenarios.