Structural Acoustic UXO Detection and Identification in Marine Environments
Abstract
This work completes the post measurement analysis of the data collected in June 2013 in the Gulf using the BOSS-40. Inclusion of AUV position data has improved image quality and the ability to correctly register buried target depths. Along-range resolution resolved the UXO long axes, and cross-range resolution was within a factor of two of the UXO diameters. In a study that included all targets, results demonstrate that acoustic color can separate the detections from the UXO/non-UXO groups. BOSS source calibration studies demonstrated that the source of AUV system does not radiate spherically causing a significant impact on false target classification performance. "Leave one out" studies indicated that training experimentally can produce classification algorithms that have a near I probability of classification for UXO targets of the type used in training and a false positive percentage of about 33% against false targets unlike the UXO from a structural acoustics point of view. While the false alarm rate is higher against targets similar to the UXO, this performance would improve significantly with the use of a more uniform source.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2016
- Accession Number
- AD1031811
Entities
People
- A. Sarkissian
- Brian H. Houston
- D. Amon
- Eric B. Williams
- H. Simpson
- Joe A. Bucaro
- M. Saniga
- T. Yoder
- Z. J. Waters
Organizations
- United States Naval Research Laboratory