Autofocus and Back-Projection in Synthetic Aperture Radar Imaging
Hyun Jeong Cho
Friday, August 26, 2016|
10:00am - 12:00pm
Add to Google Calendar
About the Event
Abstract: Spotlight-mode Synthetic Aperture Radar (SAR) imaging has received considerable attention due to its ability to produce high-resolution images of scene reflectivity. One of the main challenges in successful image recovery is the problem of defocusing, which occurs due to inaccuracies in the estimated round-trip delays of the transmitted radar pulses. The problem is most widely studied for far-field imaging scenarios with a small range of look angles since the problem formulation can be significantly simplified under the assumptions of planar wavefronts and one-dimensional defocusing. In practice, however, these assumptions are frequently violated. The main aim of this work is to devise a more broadly effective autofocus approach by adopting MultiChannel Autofocus (MCA), a subspace-based algorithm, to more realistic scenarios. By forming the solution space in a domain where the defocusing effect is truly one-dimensional, we show that drastically improved restorations can be obtained for applications with small to fairly wide ranges of look angles. When the terrain topography is known, we utilize the versatile backprojection-based imaging methods in the model formulations for MCA to accurately account for the underlying geometry. The proposed extended MCA shows reductions in RMSE of up to 50% when the underlying terrain is highly elevated. We also analyze the effects of various geometric parameters and other constraints on the quality of restored images.
Sponsor(s): Professor David C. Munson Jr.
Faculty Sponsor: Professor David C. Munson Jr.
Open to: Public