Motion Error Compensation in Airborne Stripmap Synthetic Aperture Radar

Abstract

Airborne synthetic aperture radar (SAR) requires motion compensation (MOCO) to be performed to form high resolution imageries. For wide-beam SAR systems, traditional range-dependent narrow-beam MOCO is insufficient to fully focus the images. In addition, for systems without high precision inertial navigation sensors (INS), autofocus techniques are also required. This thesis proposes two novel techniques based on the widely used Subaperture- Topography and Aperture-Dependent (SATA) algorithm. It is well known that SATA’s capability is limited in compensating high frequency motion errors. Further analyses of the algorithm show that the target resolution is traded off with the sidelobe level when a fixed set of subaperture parameters is chosen. In the novel techniques, these parameters are chosen flexibly to overcome the trade-offs in conventional SATA. By adaptively selecting the subaperture block length based on the motion error frequency and target’s range, the adaptive SATA improves the image contrast and the far range resolution by up to 10%. By combining SATA of different block sizes with the concept of complex dual apodization, the apodized SATA improves the overall target resolution and sidelobe suppression at all ranges. The algorithms are applied to a Polarimetric L-band Imaging Radar (PLIS) dataset collected in Victoria, Australia. Acombined Phase Curvature Autofocus (PCA)-Multi-aperture mapdrift (MAM) based autofocus approach is proposed. The nonparametric PCA can estimate both low and high frequency motion errors, but it cannot perform reliably in absence of dominant scatterers, as is often the case for rural scenes. The parametric MAM is suitable for a wider scene type including urban and rural, but it is limited to low frequency motion error estimation only. By combining the merits of these two approaches, the novel technique in this thesis can estimate and correct low to high frequency motion errors for a wide range of scenes. Significant improvements in image contrast and overall target focusing quality, compared to the standard mapdrift and PCA-based approaches, are achievable with the algorithm as is demonstrated with PLIS dataset. Empirical analysis performed on a set of motion compensated and uncompensated images shows a substantial reduction in the measured backscattering coefficient due to motion error. This will have potential impacts on the accuracy of the information retrieved from these imageries. By bringing notable improvements to image focusing, the proposed MOCO techniques have the potential to benefit applications based on high quality SAR imageries.