| Popis: |
In this work, we present an approach for the three dimensional (3D) parameter estimation of targets from sparse, multipass synthetic aperture radar (SAR) data. We use scattering models that allow the inverse radar scattering problem to be solved in a low dimensional space. Four radar scattering models are presented. We assume the geometric theory of diffraction which allows complex targets in the scene to be modeled as a superposition of these radar scattering models.Our approach processes data in the image domain to exploit the spatial separa¬tion of objects that is not apparent in the Fourier domain. We work with a set of two dimensional (2D) images rather than one 3D image to achieve lower complexity with no information loss. Layover is addressed which becomes inherent when working with 2D images of 3D information. The model’s parameter vector is sampled to cre¬ate templates which provide an initialization for the following parameter refinement procedure. Accurate parameter initializations are especially required in this problem due to the many local maxima and minima that lead to suboptimal parameter esti¬mates. Our approach also addresses the issue of high image sidelobes resulting from sparse and nonuniform sampling.The algorithm is applied to both synthetic data and measured data from the Air Force Research Laboratory’s GOTCHA program. Successful results are seen in the synthetic test cases and some of the test cases using the GOTCHA dataset. |
