Multi-Sensor Fusion for Buried Object Detection

Abstract

During this project, we will explore new methods for multi-resolution multi-sensor target detection, target classification, and data fusion given uncertain and imprecise groundtruth. These methods will be developed for subsurface object detection using electromagnetic in-duction (EMI) and ground penetrating radar (GPR) data. We aim to develop sensor-specific target detection and classification algorithms and, then, develop fusion methods that can address mis-registration between sensor sources and well as uncertainty and imprecision in ground-truth of training data. Sensor-specific prescreening, classification and fusion algorithms will be developed during this project. The sensor-specific algorithms will be designed to leverage the unique characteristics and capabilities of each EMl and GPR sensor being considered. During this proposed work, the following three research objectives will be addressed: I. Develop prescreening algorithms that identify potential target locations, i.e., alarms using: (a) Sensor-specific EMI and GPR approaches (b) Joint EMI-GPR approaches 2. Study and develop target classification algorithms for EMI and GPR data that appropriately account for imprecise and uncertain label information. 3. Investigate and implement alarm and decision fusion approaches that account for spatial sensor mis-registration. Throughout the study and implementation of these three research objectives, extensive evaluation and comparison to existing methods in the literature will be conducted. Regression testing will be used as improvements are made to the detection system overall to ensure previous performance is maintained or improved upon. Results will be evaluated using Receiver Operating Characteristic (ROC) curves as well as confusion matrices and other standard evaluation metrics.

Document Details

Document Type
DoD Grant Award
Publication Date
Sep 11, 2018
Source ID
W911NF1710213

Entities

People

  • Alina Zare

Organizations

  • Army Contracting Command
  • United States Army
  • University of Florida

Tags

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Military/Explosive Ordnance Disposal (EOD) Technology
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML