A belief theoretic approach for characterization of underwater munitions
Thanuka L. Wickramarathne, S. Negahdaripour, Kamal Premaratne, Lisa N. Brisson, Pierre P. Beaujean
Abstract
Thanuka L. Wickramarathne, S. Negahdaripour, Kamal Premaratne, Lisa N. Brisson, Pierre P. Beaujean
Abstract
Characterization, management and remediation of military munitions, especially in underwater environments, is a challenging task given all the technical and physical barriers. Optical cameras are better suited for identifying the physical shape of objects. But in underwater, low visibility almost prohibits the use of these cameras. Acoustic imaging is a good alternative to this, but the characteristics of imaging along with numerous artifacts of physical systems which are not easy to model, makes the object recognition task non-trivial. We explore here the possibility of exploiting the geometry of the object shadows for identification of objects itself. The inherited imperfections of the data and the numerous artifacts of sonar systems are counteracted via the use of a fusion algorithm which incorporates evidence from multiple perspectives. A Dempster-Shafer belief theoretic evidence updating scheme which is capable of modeling a wider variety of data imperfections is used for the fusion task. We illustrate the method via the use of real data obtained at a test site located in the Florida Atlantic University premises.
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Characterization, management and remediation of military munitions, especially in underwater environments, is a challenging task given all the technical and physical barriers. Optical cameras are better suited for identifying the physical shape of objects. But in underwater, low visibility almost prohibits the use of these cameras. Acoustic imaging is a good alternative to this, but the characteristics of imaging along with numerous artifacts of physical systems which are not easy to model, makes the object recognition task non-trivial. We explore here the possibility of exploiting the geometry of the object shadows for identification of objects itself. The inherited imperfections of the data and the numerous artifacts of sonar systems are counteracted via the use of a fusion algorithm which incorporates evidence from multiple perspectives. A Dempster-Shafer belief theoretic evidence updating scheme which is capable of modeling a wider variety of data imperfections is used for the fusion task. We illustrate the method via the use of real data obtained at a test site located in the Florida Atlantic University premises.
Key concepts: Sonar, Visibility, Underwater, Computer science, Task (project management), Identification (biology), Object (grammar), Sensor fusion