2-D linear predictive compression of complex synthetic aperture radar (SAR) images

Current compression of complex SAR imagery uses line-by-line 1-D linear prediction which can cause discontinuities during reconstruction between lines. This paper introduces computationally fast full 2-D linear prediction techniques which can process entire complex SAR images (or sub-images) with another 10X factor improvement in complex data compression over 1-D techniques, while yielding better weak target and shadow area preservation. 2-D linear prediction also compresses linear extent target features not possible with 1-D algorithms.