Method for Minefields Mapping by Imagery from Unmanned Aerial Vehicle

Anotace: The paper proposes a method for minefields mapping by the centimeter resolution imagery from a copter-type unmanned aerial vehicle (UAV) which is equipped with multispectral camera and thermal infrared camera. The research methodology is the probability fusion by each sensor and the subsequent decision making on the landmine presence/absence. Models for the landmine detection in multispectral and thermal images are considered. The training sample structuration is proposed for the landmine detection reliability enhancement. The local temperature anomalies of landmine size are allocated by sliding window scanning the thermal image. The experimental performance of actual landmines detection at a special test site in Ukraine is described. The probability of correct landmine detection was 0.92 while with a false alarm probability it was 0.45.