Autoregressive modeling of dechirped spotlight-mode sar rawdata in transform domain

Abstract

Raw data collected by synthetic aperture radar (SAR) is commonly assumed to be uncorrelated and with a zero-mean Gaussian distribution. In this paper, we show—both analytically and numerically—that the range-wise inverse Fourier transform of the dechirp-on-receive circular SAR data exhibits significant correlation in the azimuth direction. Moreover, we show that a block adaptive autoregressive model well represents the transformed SAR data.