In remote sensing, accurate identification of far objects, especially concealed objects is difficult. In this study, to improve object detection from a distance, the hyperspecral imaging and wideband technology are employed with the emphasis on wideband radar. As the wideband data includes a broad range of frequencies, it can reveal information about both the surface of the object and its content. Two main contributions are made in this study:1) Developing concept of return loss for target detection: Unlike typical radar detection methods which uses radar cross section to detect an object, it is possible to enhance the process of detection and identification of concealed targets using the wideband radar based on the electromagnetic characteristics conductivity, permeability, permittivity, and return loss of materials. During the identification process, collected wideband data is evaluated with information from wideband signature library which has already been built. In fact, several classes (e.g. metal, wood, etc.) and subclasses (ex. metals with high conductivity) have been defined based on their electromagnetic characteristics. Materials in a scene are then classified based on these classes. As an example, materials with high electrical conductivity can be conveniently detected. In fact, increasing relative conductivity leads to a reduction in the return loss. Therefore, metals with high conductivity (ex. copper) shows stronger radar reflections compared with metals with low conductivity (ex. stainless steel). Thus, it is possible to appropriately discriminate copper from stainless steel.2) Target recognition techniques: To detect and identify targets, several techniques have been proposed, in particular the Multi-Spectral Wideband Radar Image (MSWRI) which is able to localize and identify concealed targets. The MSWRI is based on the theory of robust capon beamformer. During identification process, information from wideband signature library is utilized. The WB signature library includes such parameters as conductivity, permeability, permittivity, and return loss at different frequencies for possible materials related to a target. In the MSWRI approach, identification procedure is performed by calculating the RLs at different selected frequencies. Based on similarity of the calculated RLs and RL from WB signature library, targets are detected and identified.Based on the simulation and experimental results, it is concluded that the MSWRI technique is a promising approach for standoff target detection.
展开▼
