Three dimensional electromagnetic sub-surface sensing by means of a multi-step SVM-based classification technique

In this work, a classification approach for three-dimensional subsurface sensing has been presented. The approach is based on a SVM-based classification algorithm for defining an occupancy map of the area under investigation. In order to deal with 3D scenarios, allowing a suitable spatial resolution without adding further measurements or processing, the algorithm has been integrated into a multi-step zooming procedure. The potentialities of the proposed approach have been assessed by means of a numerical validation with blurred synthetic data. Although preliminary, the obtained results suggest that an extension to real on-line 3D cases is worth to be pursued.