DSP seminar, Wed Sep. 07 Time and location: 4:00-5:00 PM 2269 Beckman Institute

Title: A MULTICHANNEL APPROACH TO SAR AUTOFOCUS

Speaker: Robert L. Morrison, Jr.

Graduate Student, UIUC

This is joint work with Prof. Minh Do of UIUC and Prof. David C. Munson, Jr. 
of the University of Michigan

Abstract:

The autofocus problem in synthetic aperture radar (SAR) is considered, where Fourier phase measurement errors 
result in improperly-focused imagery.
We present a new approach to the SAR autofocus problem based on ideas from blind multichannel deconvolution.
Specifically, the restored (focused) image is obtained as the solution of a linear system of equations.
Under certain conditions on the underlying image support, and in the absence of noise, the restoration is 
exact - regardless of the characteristics of the scene or the nature of the phase error.  When these conditions 
are approximate, or in the presence of additive noise, the linear framework provides an approximate solution 
subspace for the perfectly-focused image, through which the SAR image restoration process may be cast into a 
similar framework with other image restoration problems.  An attractive feature of this approach is that the 
focused image is estimated directly; existing autofocus techniques that estimate the Fourier phase error are 
subject to ambiguities up to a linear phase term, which cause a circular shift in the restored image.  
We demonstrate the proposed technique on synthetic and actual SAR scene models, and compare the performance 
with existing autofocus methods.