Statistical Classification of Buried Unexploded Ordnance Using Nonparametric Prior Models

Abstract

Abstract-We used kernel density estimation (KDE) methods to build a priori probability density functions (pdfs) for the vector of features that are used to classify unexploded ordnance items given electromagnetic-induction sensor data. This a priori information is then used to develop a new suite of estimation and classification algorithms. As opposed to the commonly used maximumlikelihood parameter estimation methods, here we employ a maximum a posteriori (MAP) estimation algorithm that makes use of KDE-generated pdfs. Similarly, we use KDE priors to develop a suite of classification schemes operating in both "feature" space as well as "signal/data" space. In terms of feature-based methods, we construct a support vector machine classifier and its extension to support M -ary classification. The KDE pdfs are also used to synthesize a MAP feature-based classifier. To address the numerical challenges associated with the optimal data-space Bayesian classifier, we have used several approximation techniques, including Laplacian approximation and generalized likelihood ratio tests employing the priors. Using both simulations and real field data, we observe a significant improvement in classification performance due to the use of the KDE-based prior models.

Gulf Petro Initiative

“…We have demonstrated that this type of classification works well even when the model assumptions (e.g. distance from transmitter) are violated (Aliamiri et al, 2007). …”

“…Comparing these pole values to a library provides a means of classifying unknown targets (Aliamiri et al, 2007). We have demonstrated that this type of classification works well even when the model assumptions (e.g.…”

“…This approach is justified by the well known low resolution of the CSEM method-particularly at the frequencies used in a marine survey. If the ultimate goal is the automated statistical classification of an unknown subsurface conductivity anomaly as a reservoir or non-reservoir, a less that accurate model will produce good results, as shown in Aliamiri, et al (2007).…”

“…To address the second problem, we drew inspiration from prior work in unexploded ordnance (UXO) detection and discrimination (Stalnaker et al, (2006), Aliamiri et al, (2007)). UXO is approximately modeled as a collection of wire loops, the electrical and geometric characteristics of which are directly related to the nature of the UXO itself.…”

A Complex Approach to UXO Discrimination: Combining Advanced EMI Forward Models and Statistical Signal Processing

“…Initially we took the time decay of the total NSMS over 25 time channels for all targets and parameterized it using the Pasion-Oldenburg law of equation (57). Taking the logarithm of that equation we arrive at the linear model…”