Improvements in GPR-SAR imaging focusing and detection capabilities of UAV-mounted GPR systems

In recent years there has been an increasing research interest in the development of airborne-based Ground Penetrating Radar (GPR) systems for safe and fast detection of buried threats such as landmines and Improvised Explosive Devices (IEDs). Compared to other sensors such as metal detectors or magnetometers, GPR is able to detect either metallic and non-metallic targets. Besides, the use of Ultra Wide Band (UWB) radiofrequency hardware allows retrieving high resolution images of the subsoil, thus improving the detection capabilities. Airborne-based GPR systems are affected by uncertainties that need to be characterized and corrected. This contribution focuses on the following three issues: (i) influence of the height information in the GPR-Synthetic Aperture Radar (GPR-SAR) images, determining which sensors provide better quality images; (ii) distortion of the imaged targets in along-track axis due to small tilting of the UWB antennas on board the UAV, proposing a coregistration technique to correct it; and (iii) revision of SVD filtering techniques, assessing a criterion to classify the different eigenimages. The analysis of these issues, as well as the validation of the proposed correction methods, are conducted using measurements taken during validation campaigns of the implemented airbornebased GPR system. For this purpose, two validation scenarios with different soil characteristics have been selected. Metallic and non-metallic targets with different shapes and sizes have been buried in them. In the case of the first scenario (wet loamy soil), detection capabilities improved from 50% (5 out of 10 targets were detected) to 80%. For the second validation scenario (loamy soil, but less humid than the first scenario), not only a 100% detection is achieved, but also the targets are clearly imaged in the GPR-SAR images obtained with the improvements presented in this contribution.