Intelligent sensor system for humanitarian demining

Abstract

In any post conflict country, landmines have become a major concern to civilians especially during the resettlement period. Biological Sensors and Metal Detectors are the most common detection technologies used commercially. One of the major drawbacks with the commercially available metal detectors is that they don’t have very good discriminating power and in practice they give very high rate of false alarms. Generally the ratio between the detection of a landmine and getting a false alarm varies in the region of 1:100 to 1:1000 depending on the location. The aim of this study is to introduce an Intelligent Discriminating System so that the false alarms would be reduced. Detailed study on Very Low Frequency (VLF) type Metal Detectors show that, the presence of a ferromagnetic object changes the phase of the signal induced in the receiving coil. This phase change heavily depends on the type of the alloy. This property is used to discriminate ferromagnetic alloys in the proposed methodology. The field survey shows that the processing speed of the system should be fast and accurate. The aliased signal from the detector receiver coil is used in this study to reduce the processing time. The received signal is further processed using Discrete Wavelet Analysis. The First Level High Frequency Sub-band signal, with Meyer Wavelet, depends highly on the type of material, sweep frequency, and the depth of the material. This processed data is used to classify the object into different classes. Modified version of Adaptive Resonance Theory (ART-1) is used in the classifying process. The results show that different metals could be classified with 5% significance, same material but different size could be classified with 10% significance and Alloys could be classified into user defined classes; this depends on the templates used in the ART- 1. Further by changing the classification algorithm, the objects could be classified into user defined groups.