Mine detection and classification using optical sensors - Initial results

There are many tactical scenarios where optical sensor systems are proving promise in detecting surface-laid mines at substantial stand-off distance. Both passive and active hyperspectral sensors have shown high capability in automatically detecting mines and other man-made objects and the capability to cue other high resolution sensors for identification. Evaluation of sensor system combinations are typically hampered by detection algorithm capability uncertainties. Information theory is being invoked in order to obtain algorithm independent comparisons of sensor suits. The emphases of this project is to obtain new ideas for optical sensing, signal processing, target and background characterisation and data and feature fusion in order to improve on mine detection and classification performance. Important phenomenology for mine detection has still to be explored, such as polarimetric properties, emissivity variation in the long wave infrared spectral region and the benefit of detailed three dimensional imaging.