EOD Buyers Guide Trade Directory of Manufacturers for all Explosive Ordnance Disposal related products, services, equipment & supplies. - Final Report Posted for Sonar Clasification of Underwater UXO

Report Posted for Sonar Classification of Underwater UXO

SERDP/ESTCP and Raymond Lim/Navy have posted a Final Report on its Personal Computer Shallow Water Acoustic Toolset (PC SWAT), software to allow sonar performance prediction with underwater UXO targets using imaging as the mode of classification. Check it out!!!

Data and Processing Tools for Sonar Classification of Underwater UXO

https://serdp-estcp.org/Program-Areas/Munitions-Response/Underwater-Environments/MR-2230/MR-2230

OBJECTIVE

This project addresses the problem of remediating underwater unexploded ordnance (UXO) using sonar to detect underwater bottom objects. Due to the presence of clutter, advanced processing algorithms are needed to distinguish UXO from clutter with a high degree of reliability. This project will enable the development and testing of these algorithms for UXO by creating a database of scattering signatures for UXO through measurements and modeling and using this database as a basis for both algorithm testing and identification of robust classification-enhancing phenomena through physics-based analysis.

SERDP has funded the Naval Surface Warfare Center – Panama City Division (NSWC PCD) in past years to expand its Personal Computer Shallow Water Acoustic Toolset (PC SWAT) software to allow sonar performance prediction with underwater UXO targets using imaging as the mode of classification. While imaging alone is capable of providing reliable classification against the larger UXO targets, this will be more difficult for small and buried UXOs. This project will complement image-based approaches to enhance the overall performance of sonar against UXO that may be distributed in bottom configurations where imaging is difficult; e.g., when significantly buried, distributed in non-uniform environments, masked by clutter.

Technical Approach

This project follows up on SERDP project MR-1666 that leveraged on-going Navy-sponsored sonar measurements and analysis to validate sonar performance models configured for UXO applications and to develop and evaluate algorithms for separating UXO from bottom clutter. While the results of that effort appear promising, it is recognized that good results from standard classification algorithms trained with limited data can be misleading since robustness cannot be guaranteed. Robust classifier performance requires training with data that samples targets and clutter under a sufficient variety of environmental conditions to create regions in feature space representing the targets and clutter that are effectively disjoint for any test set. Since the collection of enough data under the variety of environmental conditions needed for effective algorithm training can be costly and time-consuming, the focus in the following years will be on ways to build a database of UXO target signatures for training and testing purposes that minimizes the associated burden of cost and time. A carefully planned set of field measurements on UXO leveraging Office of Naval Research (ONR) tests will be utilized. Finite element (FE) modeling of target signatures will continue to be emphasized. Using techniques developed under ONR funding, UXO-like targets will be synthetically embedded into new environments. Investigations on the scattering phenomenology will be performed on the data collected or simulated to identify signal effects that provide a unique capability to classify targets. Finally, assessments of classification performance will leverage Navy sponsored tests of advanced classifier algorithms using the database created.

Benefits

One benefit of this work will be cost savings from reduction of the number or duration of field tests of specific sonar designs. Classifier performance is one of the critical factors in assessing overall performance of an underwater sonar system. It is also one of the most difficult components of overall sonar performance to quantify. By creating a comprehensive database of target and clutter signals for classifier training and testing, the Department of Defense can produce better assessments of overall performance of sonar designs. Furthermore, robust classification and identification algorithms will make sonar a highly effective, rapid underwater inspection tool. Development of high-fidelity sensor simulations for UXO remediation requires understanding the fundamental acoustic propagation and scattering processes in shallow-water environments. Increased knowledge of these processes will be of great benefit to the scientific community. (Anticipated Project Completion - 2014)