Accurate estimation of stitching error for high-resolution multi-sub-bands SAR system with image sharpness analysis

Multi-sub-bands stitching technique is an effective method to achieve ultra-high range resolution in synthetic aperture radar (SAR) system. However, sub-bands stitching error results in artefacts in multisub- bands SAR images, e.g. defocused targets, higher sidelobes, which seriously degrade SAR image quality and cannot be mitigated by weighting function. A novel stitching error estimation method was presented. The stitching error is modelled by regarding phase error as function of range frequency, which will be estimated and compensated to reach the maximum image sharpness. First, amplitude balance is applied to single look complex SAR image in range frequency domain. Subsequently, based on image sharpness, an iterative optimisation algorithm is designed to estimate and compensate the stitching phase error. Finally, experimental results using both real RADARSAT-2 and airborne SAR dataset validate proposed method perform well in arbitrary scenes, and it outperforms the existing methods.