Extraction of weak target features from radar tomographic imagery

Radar Tomography is the process of 3D reconstruction of a measurement domain using a multistatic distribution of transmitters and receivers. Geometric diversity of these elements increases the information contained in the measurements. The process of determining the permittivity and conductivity profile of the measurement domain, and therefore the shape of the target, from the scattered field measurements is an inverse problem. This is solved using principles of linear scattering (Born approximation), which lead to a linear relationship between the measured returns and the target shape. One limitation of radar tomography is that strong scatterer sidelobes in the measurement domain can interfere with the echoes from weak scatterers, decreasing the system's ability to detect certain target feature. In this paper, we propose a method to increase overall image quality by modelling the strong scatterers in the measurement domain as dipoles which behave as secondary transmitters. The purpose of this model is to reduce the effects of the sidelobes from the strong scatterers. We estimate the electromagnetic characteristics for each dipole in the model by representing the cells in the measurement domain's image as dyadic functions. The eigenvalue and eigenvector for each cell represents phase and magnitude for the modelled dipole. The process of modelling targets as dipoles can be repeatedly applied, addressing one strong scatterer at a time, to decrease uncertainty in the measurement domain. Simulations and results demonstrate this concept.