This project presents a development of a small radar system for object detection (including landmines) using advanced radar concepts and implementing advanced imaging concepts using multiple views of the examined area. Existing concepts using radar techniques are based on generating a pulse with a short duration in time domain or generating a series of signals with different frequencies and using signal processing techniques to reconstruct the time domain signal. Both approaches can be efficiently implemented using telecommunication components, since today’s Ground Penetrating Radars (GPR) operate in the frequency range from 50MHz to 3GHz.
GPR data are commonly acquired using a perpendicular broadside antenna configuration, with the transmit and receive antennas are oriented parallel to each other and perpendicular to the direction of survey line. As most GPR systems employ linearly polarized dipole antennas, the transmit antenna emits an electromagnetic (EM) wave whose electric field is polarized parallel to the long axis of the dipole, and the receive antenna records only the component parallel to its long axis. However, it has been noted that various targets of GPR surveys, such as buried pipes and fractures, have polarization-dependent scattering characteristics. Therefore, polarization dependent scattering properties have important implications for target detection, survey design, and data interpretation. Polarimetric technology has been one of the most important advances in microwave remote sensing during the recent decades. Currently, polarimetric technology is being introduced into GPR to improve the detection capability. A full-polarimetric borehole radar system has been developed with combinations of dipole antennas and axial slot antennas. Because dipole antenna radiates vertical (V) electric field, and axial slot antenna radiates horizontal (H) electrical field, the system can acquire fully polarimetric data sets, which are used to analyse the subsurface fracture characterization, in the drilled borehole.
Most of the experiments in the literature are very new and have not been applied to the landmine detection. The project will exploit possibilities of polarimetry for landmine detection. Therefore, one of the goals will be to develop a small radar that uses all 4 polarimetry channels to exploit polarimetry signatures of the landmines.
The GPR covers area illuminated by antenna pattern, that is usually up to 10 cm wide, therefore radar platform must move repeatedly over the same area to detect objects under investigation. The recent advances in Multiple Input Multiple Output (MIMO) radars showed that larger area with higher resolution can be detected.
Different GPR imaging methods have been used in the past. Recently, SAR and microwave tomography signal processing have attracted lot of interest. To obtain an image using the microwave tomography, multiple emitters and sensors are usually employed. The inverse scattering method is used to evaluate the dielectric permittivity from the measured electric field. This is a highly non-linear and ill-posed problem. Therefore, usually, linearization using the Born approximation is used. The problem is solved in frequency domain and reduces to finding an inverse matrix after the discretization. Different antenna polarizations can be used. To use microwave tomography, multiple views of the examined subsurface are required. We propose to use multiple UAV, carrying transmitters and receive sensors to perform measurements of the scattered signal. We will use two scenarios, the multi-monostatic, where the transmitter and receiver are placed on the same UAV and the multi-bistatic with a stationary transmitter at a given position and a receiver moving around it, followed by a change of the transmitter position, when the whole procedure repeats.
The Artificial Intelligence will be used to detect objects within the polarimetric data. The complex valued deep convolutional neural networks (CV CNN) showed very promising results in feature extraction of polarimetric data. Within this project CNN will be used to detect objects in radar data.
The objectives of the proposal are:
- To develop a small size GPR that is convenient for attaching to small vehicles or UAV
- To develop Multiple Input Multiple Output radar that will provide a higher accuracy and will use polarimetric features of the targets
- To develop advanced SAR and microwave tomography imaging techniques using the multi-monostatic mode (single drone) or multi-bistatic mode (moving receive sensor).
- To use Artificial Intelligence to detect and classify landmine, UXO and other explosives
English 
Slovenian