Characterisation of dielectric obstacles using ultra-wideband techniques

In this paper ultra-wideband techniques are utilized for the classification and localisation of small dielectric objects with different permittivities, which may be considered as foreign bodies in otherwise quasi homogeneous dielectric layers. The dielectric obstacles are irradiated with antennas using transmission techniques. Pulses are applied having a width of approximately 400 ps. The signals are sampled after interfering with the object under test and processed using principal component analysis and artificial neural networks. Measurements are carried out in order to estimate the obtained spatial resolution and predict the material properties. The results show that the localisation is independent from the dielectric classification. The spatial resolution is better than the involved wavelengths anticipate.