• geophysics
• engineering
• air-coupled GPR
• IED

The following master's thesis aims to analyze and test algorithms, methodologies and systems for Ground penetrating Radar (GPR) applications. The GPR has been shown to be a useful tool for Non-Destructive Testing (NDT) that has been successfully introduced in various sectors such as geology, military applications, civil engineering, and archeology. Thanks to the GPR it is possible to detect buried objects without coming into direct contact with them.
In this thesis, several existing techniques for the processing of data (measurements) acquired with a GPR have been studied, in order to determine the advantages and limitations of each:
- Delay and Sum (DAS), is a technique that is based on beam propagation in a medium, allowing the detection of reflections that occur due to the discontinuity between two media. In this master's thesis it has been used for monolayer media, although with the appropriate modifications it can also be used in the case of multiple layers.
- Inverse electromagnetism techniques, which are based on electromagnetic field equations to relate the measured fields to objects. These techniques are mainly based on a minimization of the cost function that relates the theoretical and experimentally observed electromagnetic fields (or, in other words: the cost function provides a measure of the fit between a parameterized curve and the data). In this master's thesis, the minimization of the cost function has been carried out by means of the Tikhonov Regularization, which is a method used in case of badly conditioned problems and instability of inverse problems, and also with the Conjugate Gradient method, which it is an iterative algorithm used for the numerical solution of systems of linear equations (with symmetric and positive matrix). Reverse electromagnetism techniques require less information in order to accurately reconstruct the geometry of an object (for example, they can work with monochrome data, that is, data acquired at a single frequency). However, they carry a higher computational cost, unlike the DAS technique.
The validation of these techniques was carried out mainly with data based on electromagnetic simulation using the gprMax software and Matlab ®, as well as with measurements using an Ultra Wide Band (UWB) radar. In the case of measurements, a portable scanner acquisition system was used, where measurements have been georeferenced using a centimeter-level precision GNSS system (to avoid the need for encoders or robotic systems, which imply higher cost and complexity). Different types of antennas were analyzed to test their impact on the detection and imaging capabilities of the proposed system. Objects placed on the ground, as well as buried objects, were tested in order to assess the ability of the techniques described above to detect and identify such objects. Thanks to the use of SAR processing algorithms, an improvement in image resolution was obtained, mitigating the characteristic hyperbolic shape of crude measurements for electrically small objects. On the other hand, it was observed that the conjugate gradient provides images with higher contrast or sharpness of the objects (regardless of whether they are located on the ground or buried) especially in the simulated scenarios with gprMax post-processed with Matlab ®. However, it has also been observed that both the Conjugate Gradient and the inversion with Tikhonov Regularization results in the amplitudes of the objects (both buried and not buried) being less than that obtained with the DAS algorithm, which it can make it difficult to detect objects.