Objective

Electromagnetic induction (EMI) data from subsurface targets must be processed and inverted in order to discriminate innocuous scrap metal or geology from UXO. At highly contaminated sites, processing data containing overlapping responses from unexploded ordnance (UXO) and clutter is a difficult task both theoretically and computationally. The total EMI response from these densely populated sites is a combination of responses from single or multiple UXO and obscuring clutter resulting in confused or confounded inversion routines. This data contamination often leads to misclassification at best, or a complete failure of the routine at worst.

The objective of this project is to mitigate the effects of dispersed metallic clutter, to resolve the contribution of independent or coupled objects to the composite EMI data, and to discriminate each object of interest based on its response under more realistic field conditions.

Specific objectives include:

1. Develop an N-target locater that, without any computationally expensive optimizations, provides an estimate of the number of objects within a sensor’s field of view and their locations and orientations. A spatial particle filter and a formulation of the HAP method for multiple targets will be investigated.
2. Formulate robust classifiers that segregate N objects into UXO and non-UXO, based on their isolated EMI responses.
3. Discriminate UXO-like targets using rigorous models that explicitly include coupling between targets if required (Normalized Surface Magnetic Source [NSMS] and Standardized Excitations Approach [SEA]).