Signal processing issues in reflection tomography

This dissertation focuses on signal modeling and processing issues of the following problems in reflection tomography: synthetic aperture radar (SAR) imaging of a runway and surroundings from an aircraft approaching for landing, acoustic imaging of objects buried in soil, and lidar imaging of underwater objects.; The highly squinted geometry of runway imaging necessitates the incorporation of wavefront curvature into the signal model. We investigate the feasibility of using the wavenumber-domain (ω - k) SAR inversion algorithm, which models the actual curvature of the wavefront, for runway imaging. We demonstrate the aberrations that the algorithm can produce when the squint angle is close to 90° and show that high-quality reconstruction is still possible provided that the interpolation is performed accurately enough, which can be achieved by increasing the temporal sampling rate. We compare the performance with that of a more general inversion method (GIM) that solves the measurement equation directly. The performances of both methods are comparable in the noise-free case. Being inherently robust to noise, GIM produces superior results in the noisy case. We also present a solution to the left-right ambiguity of runway imaging using interferometric processing.; In imaging of objects buried in soil, we pursue an acoustic approach primarily for detection and imaging of cultural artifacts. We have developed a mathematical model and associated computer software in order to simulate the signals acquired by the actual experimental system, and a bistatic SAR-type algorithm for reconstruction. In the reconstructions from simulated data, objects were detectable, but near-field objects suffered from shifts and smears. To account for wavefront curvature, we formulated processing of the simulated data using the 3-D version of the monostatic ω - k algorithm.; In lidar imaging of underwater objects, we formulate the problem as a 3-D tomographic reconstruction problem. We have developed software to simulate lidar returns at airborne receivers using the bistatic lidar return equations. Our simulator can model multiple scattering and absorption for various water types and system parameters. Our simulated data fits the characteristics of real data very well. We present our reconstruction results from the simulated and real data, and comparatively discuss the reconstructions.