Vector Radar Wave Diffraction Tomography Maximum Likelihood Estimation

Signal processing algorithms for target detection have been developed for ground penetrating radar within the theory of maximum likelihood estimation (DTMLE). Diffraction tomography is a wave-based imaging method whereby received signals are backpropagated into the supporting medium to map the spatial variations in the properties of inhomogeneities. When mated with a maximum likelihood estimator, this approach yields an “image” of the spatial variations in the log likelihood function; the likelihood of a specified target occurring at a particular location. This effort is an extension of previous work on scalar wave DTMLE to the case of vector waves. In addition, the analysis is extended into the evanescent spectrum to exploit the greater spatial resolution available at these shorter length scales (higher spatial frequencies). The motivating application for these algorithms is tunnel detection. The use of vector wave-based algorithms allows the explicit consideration of the three dimensionality of targets, in particular the large aspect ratio of tunnels, from multiple one dimensional measurements employing various antenna polarizations. The analysis of simulated data shows that a model tunnel (a cylinder) can be differentiated from a spherical target having the same cross-sectional area. Confirmatory field studies over a known tunnel demonstrate that this tunnel could be unambiguously detected and accurately located. The inclusion of evanescent wave allows this detection even at low frequency (25 MHz) and the use of three different antenna polarizations provides sufficient information to quantify the orientation of the tunnel axis.