Algorithmic classification of active/inactive unexploded ordnance using neutron analysis data

For over a decade, the Applied Physics Institute at Western Kentucky University has been assisting the U.S. Army to determine the active/inactive status of unexploded ordnance (UXO) using a Pulsed Fast/Thermal Neutron Analysis (PFTNA) technique that indicates the elemental composition of the test item. Typically, their methods require a skilled analyst to interpret the spectral data collected. The author has developed a wavelet-based algorithm that automatically interprets the spectral data and makes a determination as to the substance being tested. The basis being used is a smooth scaling vector and multiwavelet developed by the author with approximation order 4, along with a prefilter developed by the author. Preliminary results from its usage on PFTNA data in UXO determination applications are described. (author)