Continuous phase corrections applied to SAR imagery

Phase error compensation is typically applied identically to every pixel in a Synthetic Aperture Radar (SAR) image. For certain modern systems and applications, this methodology is on the verge of becoming insufficient. We present Pixel-Unique Phase Adjustment (PUPA), an algorithm that performs an arbitrary spatially varying correction. We treat this as a deconvolution problem for which the goal is to minimize the cost function corresponding to the maximum likelihood estimate of the restored image. Our approach uses an iterative, gradient-based optimization algorithm. This method handles nonparametric phase errors and removes distortions exactly. We present results on real SAR data and demonstrate that quality is limited only by measurement noise. We analyze performance in terms of both computational complexity and memory requirements, and discuss two different implementations that allow a tradeoff to be made between these resources.