What can be expected from Computerised Image Analysis Techniques for Airborne Minefield Detection?

We investigate the applicability of image analysis to the problem of minefield detection using airborne remote sensing images. We have applied different image processing methods for the detection of minefields, minefield indicators and individual mines. The proposed algorithms involve the extraction of linear features (roads, paths, fences, wires,...), detection of periodic patterns (e.g. regular minefields, regularly placed minefield indicators) and segmentation of the image in regions of interest. Different scales of airborne images (1/500 and 1/2000) have been investigated. The algorithms have been applied on images of a test field in Belgium and real minefields in Mozambique. Introduction Minefield detection is usually carried out using shortrange ground-based sensors. These are sometimes fairly effective, but without large-area coverage capability, which makes the whole process very slow. On the other hand, recent developments in airborne remote sensing open new perspectives for ..