Detection of low-metal content objects by evidential fusion of mine detection sensors

The metal detector is the most widely used sensor in humanitarian demining. However, it does not provide enough information for the detection of objects with a low metal content. In this paper, we propose to combine it with other sensors, as investigated within HUDEM, a technology exploration project on humanitarian demining. Based on the specificities of such objects, we propose a method for modelling and fusion of measures extracted from the sensors in the framework of the DempsterShafer theory. We show how these objects can be distinguished from non-metallic and highly metallic ones. We also discuss how the proposed method deals with situations where the sensors do not detect the same object.