Kalman Filter Disciplined Phase Gradient Autofocus for Stripmap SAR
Yake Li, Siu O’Young
Abstract
The phase gradient autofocus (PGA) and its improvements have been aimed to estimate the phase error exclusively from the phase of raw data. In this article, we introduced the Kalman filter (KF) into stripmap PGA (or phase curvature autofocus) by taking advantage of the continuous movement of the aircraft. The fundamental principle is to build a kinematic model and a measurement model to predict the phase curvature of the next subaperture, and to correct the measurement (phase curvature) of the current subaperture. The advantages of employing KF are as follows: 1) the inaccurate PGA estimation due to wrong target selection, serious phase error, or low signal-to-clutter ratio can be corrected by a well-maintained KF; 2) the prediction of the KF can be applied to the data of the next subaperture before phase estimation, to decrease the algorithm converge time, and to increase the estimation accuracy; and 3) KF disciplined PGA naturally fits the sequential processing needs and is capable of generating good phase error estimation in one execution. This helps real-time synthetic aperture radar (SAR) autofocus and motion compensation. The disciplining of the autofocus using KF is not restricted to PGA-based algorithm. It can be applied to other subaperture-based autofocus algorithms.