Optical and Magnetometric Data Integration for Landmine Detection with UAV

The joint processing of optical imagery and signals from an onboard fluxgate magnetometer for landmine detection is described in this paper. The basic sensors carried by unmanned aerial vehicles (UAV) enable remote landmine detection, improving the safety of demining. The general methodology for processing both optical and magnetometric data is described. Modern machine learning (ML) and deep learning (DL) techniques are engaged for landmine detection; in particular, optical images are analyzed by a convolutional neural network (CNN), while statistical anomalies are extracted from magnetometer signals. Data integration is performed at the optical and magnetometric detection results level using the Bayesian probabilistic rule. The combination of an optical camera and a magnetometer provides significant reliability enhancement in unburied landmine detection. The proposed methodology will be quite useful for the humanitarian demining of a wide area, improving the reliability of data obtained by remote sensing methods, thus accelerating wide area exploration.