Parameterisation of non-homogeneities in buried object detection by means of thermography

Abstract

Landmines and their natural environment form a system of complex dynamics with variable characteristics. A manifestation of that complexity within the context of thermography-based landmines detection is excessive noise in thermograms. That has severely suppressed application of thermography in landmines detection for the purposes of humanitarian demining. (To be differentiated from military demining and demining for military operations other than war [Land Mine Detection DOD's Research Program Needs a Comprehensive Evaluation Strategy, US GAO Report, GAO-01 239, 2001; International Mine Action Standards, Chapter 4.––Glossary. Available at: < http://www.mineactionstandards.org/IMAS_archive/Final/04.10.pdf>].) The discrepancy between the existing role and the actual potential of thermography in humanitarian demining motivated systematic approach to sources of noise in thermograms of buried objects. These sources are variations in mine orientation relative to soil normal, which modify the shape of mine signature on thermograms, as well as non-homogeneities in soil and vegetation layer above the mine, which modify the overall quality of thermograms. This paper analyses the influence of variable mines, and more generally the influence of axially symmetric buried object orientation on the quality of its signature on thermograms. The following two angles have been extracted to serve as parameters describing variation in orientation: (i) θ––angle between the local vertical axis and mine symmetry axis and (ii) ψ––angle between local vertical axis and soil surface normal. Their influence is compared to the influence of (iii) d––the object depth change, which serves as control parameter. The influences are quantified and ranked within a statistically planned experiment. The analysis has proved that among the parameters listed, the most influential one is statistical interaction dψ, followed with the statistical interaction dθ. According to statistical tests, these two combinations are considered the most significant influences. The results show that the currently applied analysis of thermography in humanitarian demining must be broadened by the inclusion of the variations in mine orientation, otherwise a decrease in the probability of mine detection, due to the presence of a systematic error, occurs.