New approach of signal processing for classification problems using a-priori information

Metal detectors based on the eddy current principle are widely used in humanitarian demining, but their main disadvantage is the high false alarm rate caused by harmless clutter objects or "uncooperative" soils. In most cases, the types of mines in the field are known a-priori. So their influence on the metal detector can be measured and stored in a database. In this paper a new method is described that uses this information to adapt the soil compensation and the signal classification to achieve a significant reduction of false alarms. Therefore a fuzzy classifier is used that adapt to the signal quality and feature variance of the known objects. The ability to lower the false alarm rate was verified by trials on the test lane for mine detection provided by the Joint Research Center in Ispra (Italy).