TITLE:    SAR Imaging of a 3-D Scene Using Spectral Estimation Techniques 


SPEAKER:  Shu Xiao
          Coordinated Science Lab
          UIUC

DATE:     Monday, November 24, 1997
TIME:     2:00 p.m.
ROOM:     B02 CSRL


ABSTRACT

We consider the problem of synthetic aperture radar (SAR) imaging of
3-D targets, given a limited amount of measured data. We assume that
the complex reflectivity function is nonzero at a maximum of M
z-coordinates for each x-y pair. We explore an image reconstruction
method proposed by Webb and Munson. This approach employs
interpolation and a series of 2-D inverse FFTs to cast the 3-D imaging
problem into the form of a high-resolution spectral estimation problem
in the z direction, at each x-y location. The series of 1-D spectral
estimation problems can be  solved using a variety of methods. We give
overviews of an eigendecomposition method, ESPRIT, and a FFT-based
method, RELAX. We apply these methods to simulated 3-D data for a
maximum of M=2 and M=4 elevations at each x-y position. We also
compare the performance of the  image reconstruction algorithms with
the Cramer-Rao lower bound. The  approach studied here provides a
direct method for 3-D radar imaging from limited SAR data, which
avoids some of the problems associated with stereo and interferometry.