Ultra-wideband synthetic aperture radar tomography compensation algorithms and noise reduction techniques

<p>Presented in this dissertation is the exploration and formulation of an Ultra-Wideband Synthetic aperture RADAR Tomographic algorithm. Specifically, the problems that are addressed relate to the coherency in which data is collected as well as limitations in the depth component. Measurements are performed using an Inverse SAR arrangement utilizing quazi-monostatic antennae. Another name for this method of imaging is Coherent Doppler Tomography. This data is then displaced from what would otherwise be considered ‘ideal’ to better emulate an environment that contains disruption, such as data obtained in high winds. The primary goal is to best emulate and track these shifts to generate an approach that caters for them. Estimation of displacement then leads into compensation for images that contain scatterers either above or below the image plane.</p> <p>Characterization of noise in respect to its effect on the resultant tomographic image must be completed before any algorithmic approach can be taken. Coherency in image integrity is governed by a set of geometric coordinates that include an angular component and a down-range. Source antenna motion in all three Cartesian directions will affect the image reconstruction differently. Therefore, the analysis that has been performed looks at in what direction is the magnitude of change greatest and to what effect they have on the image map.</p> <p>A step-by-step explanation of a new algorithm utilizing Point Scatterer Vector Tracking of an isotropic scatterer is demonstrated. Inspired by the Kalman filter (similar to what is used in GPS), the algorithm uses previous peak position as well as a defined path estimation to record the peak location at every angular sample. The method is demonstrated using both a single target and multiple targets along with an accompanying linear regression model to prove the effectiveness.</p> <p>Furthermore, a new compensatory algorithm to aid in the planar nature of a CDT image is explored. A tomographic image depicts what would be a two-dimensional slice of a target. This indicates that when prominent scatterers are located either above or below the image reference plane, divergence in the Point Spread Function leads to a degree of blurring. A new algorithm using a technique of Dual Layer Peak Matching has been formulated. It measures the divergences in several reference scatterers placed within the image plane and estimates the contouring of the structures surface, leading to increased coherency.</p> <p>Exploration into the limitations of CDT with respect to both displacement noise and depth has enabled the possibility of reconstruction. The result is a tomographic image that can be used to best estimate the electromagnetic signature of a target no matter if the data was collected in a controlled environment or a hostile one.</p>