TY - JOUR
T1 - A comparison of the performance of statistical and fuzzy algorithms for unexploded ordnance detection
AU - Collins, Leslie M.
AU - Zhang, Yan
AU - Li, Jing
AU - Wang, Hua
AU - Carin, Lawrence
AU - Hart, Scan J.
AU - Rose-Pehrsson, Susan L.
AU - Nelson, Herbert H.
AU - McDonald, Jim R.
N1 - Generated from Scopus record by KAUST IRTS on 2021-02-09
PY - 2001/2/1
Y1 - 2001/2/1
N2 - In most field environments, unexploded ordnance (UXO) items are found among extensive surface and subsurface clutter and shrapnel from ordnance. Traditional algorithms for UXO remediation experience severe difficulty distinguishing buried targets from anthropic clutter. Furthermore, naturally occurring magnetic geologic noise often adds to the complexity of the discrimination task. These problems render site remediation a very slow, labor-intensive, and inefficient process. While sensors have improved significantly over the past several years in their ability to detect conducting and/or permeable targets, reduction of the false alarm rate has proven to be a significantly more challenging problem. Our work has focused on the development of signal processing algorithms that incorporate the underlying physics characteristic of the sensor and of the anticipated UXO target, in order to address the false alarm issue. In this paper, we describe several algorithms for discriminating targets from clutter that have been applied to data obtained with the multisensor towed array detection system (MTADS). This sensor suite has been developed by the U.S. Naval Research Laboratory (NRL), and includes both electromagnetic induction (EMI) and magnetometer sensors. We describe four signal processing techniques that incorporate features derived from simple physics-based sensor models: a generalized likelihood ratio technique, a maximum likelihood estimation-based clustering algorithm, a probabilistic neural network, and a subtractive fuzzy clustering technique. These algorithms have been applied to the data measured by MTADS in a magnetically clean test pit and at a field demonstration. We show that overall the subtractive fuzzy technique performs better than the alternative techniques when the training and testing data sets are separate. The results also allow us to quantify the utility of fusing the magnetometer and the EMI data, and we show that performance is improved when both EMI and magnetometer features are utilized. The results indicate that the application of advanced signal processing algorithms could provide up to a factor of two reduction in false alarm probability for the UXO detection problem.
AB - In most field environments, unexploded ordnance (UXO) items are found among extensive surface and subsurface clutter and shrapnel from ordnance. Traditional algorithms for UXO remediation experience severe difficulty distinguishing buried targets from anthropic clutter. Furthermore, naturally occurring magnetic geologic noise often adds to the complexity of the discrimination task. These problems render site remediation a very slow, labor-intensive, and inefficient process. While sensors have improved significantly over the past several years in their ability to detect conducting and/or permeable targets, reduction of the false alarm rate has proven to be a significantly more challenging problem. Our work has focused on the development of signal processing algorithms that incorporate the underlying physics characteristic of the sensor and of the anticipated UXO target, in order to address the false alarm issue. In this paper, we describe several algorithms for discriminating targets from clutter that have been applied to data obtained with the multisensor towed array detection system (MTADS). This sensor suite has been developed by the U.S. Naval Research Laboratory (NRL), and includes both electromagnetic induction (EMI) and magnetometer sensors. We describe four signal processing techniques that incorporate features derived from simple physics-based sensor models: a generalized likelihood ratio technique, a maximum likelihood estimation-based clustering algorithm, a probabilistic neural network, and a subtractive fuzzy clustering technique. These algorithms have been applied to the data measured by MTADS in a magnetically clean test pit and at a field demonstration. We show that overall the subtractive fuzzy technique performs better than the alternative techniques when the training and testing data sets are separate. The results also allow us to quantify the utility of fusing the magnetometer and the EMI data, and we show that performance is improved when both EMI and magnetometer features are utilized. The results indicate that the application of advanced signal processing algorithms could provide up to a factor of two reduction in false alarm probability for the UXO detection problem.
UR - http://ieeexplore.ieee.org/document/917111/
UR - http://www.scopus.com/inward/record.url?scp=0035245390&partnerID=8YFLogxK
U2 - 10.1109/91.917111
DO - 10.1109/91.917111
M3 - Article
SN - 1063-6706
VL - 9
SP - 17
EP - 30
JO - IEEE Transactions on Fuzzy Systems
JF - IEEE Transactions on Fuzzy Systems
IS - 1
ER -