A Kalman filter-based approach to detect landmines from metal detector data

Abstract

Metal detectors play a significant role in landmine detection. Automatic sensor fusion is required to improve the performance of ground penetrating radar (GPR)-metal detector multi-sensor systems. The existing version of the Kalman filter-based detection algorithm has been adapted for automatic detection and discrimination of landmines in metal detector data. In this algorithm, multi-channel metal detector output data are fused to produce a distribution of probabilities of the presence or absence of a target. Performance of this algorithm has been assessed using data obtained by burying a number of simulant landmines, canonical targets and shrapnel in different soil types.