Through-the-Wall Target Detection by Energy Features Extracted from Simulated Ultra Wideband Radar Signals

The ultra-wideband ground penetrating radars (GPRs) are successful in sensing not only conductors but also dielectric objects. For that reason, GPRs are used in many applications to detect and identify objects, which are hidden behind obstacles such as walls or buried under ground. In such problems, the early reflections from the wall or ground surfaces are very strong. These strong electromagnetic wave components must be removed during the preprocessing phase by suitable and effective methods as they lead to detection/identification errors by masking the relatively weak signals reflected/scattered from the actual targets. The novel and effective method implemented in this work is also fast due to its low computational cost and it is used to obtain an energy based feature curve corresponding to a time signal measured by the radar. The details of feature curves around the regions with very large slopes indicate the presence and location of not only wall/ground surfaces but also buried/hidden objects with enhanced accuracy at a single computational step. In this study, we firstly simulate the GPR data for four different spherical targets of the same size, which are composed of different materials over the frequency band from 0.2 GHz to 7.5 GHz. Each sphere is placed behind the same concrete wall at the same distance, one at a time. Then, the energy based feature curves are computed for the simulated GPR data to investigate their potential benefits in the problem of through-the-wall target detection/identification.