Target detection in unfocused SAR/ISAR images: A geometric approach

In synthetic aperture radar (SAR) and inverse synthetic aperture radar (ISAR), targets with uncompensated motion are blurred in the range-Doppler radar image. Our goal with this paper is twofold: (1) to analyze the image formation process in SAR/ISAR and develop a model for the distortion introduced by inadequate motion compensation; (2) to design a detection scheme that is robust to the smearing effects caused by inadequate motion compensation. We show that the blurring is equivalently modeled by describing the uncompensated target image as the superposition of space shifted echoes of a fully compensated target image. We then extend to the two-dimensional (2D) model, a geometrically based robust detection scheme we developed for the one-dimensional (1D) case. In this paper, we describe the structure and the design of the 2D robust detector in detail