Metal object localisation and discrimination by extended symmetry and phase features

Handheld metal detectors are one of the most common devices used for clearing landmines in the context of humanitarian demining. However, their capability to discriminate between different objects is limited, leading to a high false alarm rate. In this work, data sets from spatially referenced measurements with an induction detector are evaluated. First, regions of interest are identified by a new method that considers signal symmetry in combination with classical threshold based methods. Second, a new approach of applying phase plots to discriminate buried objects is presented that applies to multi-frequency continuous wave metal detectors. Multidimensional phase observables are constructed that allow successful subsequent classification. The capability of the new methods is demonstrated using data sets from numerical simulations as well as from different test lanes. The achieved improvements in the field of object discrimination are presented.