Multiple advances in diverse fields are expected to transition synthetic-aperture imaging technology from a dedicated single platform requiring an operator to a large number of small platforms operating autonomously. Such a swarm of sensors is expected to provide orders-of-magnitude performance gains relative to a single, dedicated platform. They are also expected to operate in complex environments involving dynamically changing scenes and multiple scattering. Such multi-platform synthetic-aperture imaging systems pose a number of challenges to image formation in addition to those involving wave propagation in complex environments. First, the requirement for scalability implies that the computational resources at each platform are limited and that moreover there may not be perfect phase coherency between platforms. This means the reconstruction algorithms have to be fast, decentralized and be able to handle phase errors. Second, the autonomy of the platforms implies non-ideal conditions from the perspective of image reconstruction: the sensors may be traversing arbitrary trajectories, and transmitting varying waveforms, etc. These challenges rule out the use of standard tomographic methods. This research involves developing theoretical foundations and corresponding constructive algorithms to address the challenges of multi-platform synthetic-aperture imaging. The fundamental developments of this project are applicable to all scattered-field-based synthetic-aperture imaging modalities, including RF and acoustics. Central to the project is microlocal analysis. This theory leads to powerful Generalized Filtered-BackProjection (GFBP) techniques that can accommodate non-ideal imaging conditions and wave propagation models for complex environments. This project investigates innovative extensions to microlocal techniques and integrates them with inverse scattering theory and statistical estimation and detection theory. Specifically, the investigators study GFBP algorithms for imaging in dynamically changing and multiple-scattering environments; analytic autofocus methods and fast GFBP algorithms.

Agency
National Science Foundation (NSF)
Institute
Division of Computer and Communication Foundations (CCF)
Application #
0830672
Program Officer
John Cozzens
Project Start
Project End
Budget Start
2008-09-01
Budget End
2013-08-31
Support Year
Fiscal Year
2008
Total Cost
$527,107
Indirect Cost
Name
Rensselaer Polytechnic Institute
Department
Type
DUNS #
City
Troy
State
NY
Country
United States
Zip Code
12180