People who attend to the problem of underwater port protection usually use sonar based systems. Recently it has been shown that integrating a sonar system with an auxiliary array of magnetic sensors can improve the effectiveness of the intruder detection system. One of the major issues that arise from the integrated magnetic and acoustic system is the interpretation of the magnetic signals coming from the sensors. In this paper a machine learning approach is explored for the detection of divers or, in general, of underwater magnetic sources that should ultimately support an automatic detection system which currently requires a human online monitoring or an offline signal processing. The research proposed here, by means of a windowing of the signals, uses Support Vector Machines for classification, as tool for the detection problem. Empirical results show the effectiveness of the method.
A Preliminary Study on SVM based Analysis of Underwater Magnetic Signals for Port Protection
GASTALDO, PAOLO;ZUNINO, RODOLFO
2009-01-01
Abstract
People who attend to the problem of underwater port protection usually use sonar based systems. Recently it has been shown that integrating a sonar system with an auxiliary array of magnetic sensors can improve the effectiveness of the intruder detection system. One of the major issues that arise from the integrated magnetic and acoustic system is the interpretation of the magnetic signals coming from the sensors. In this paper a machine learning approach is explored for the detection of divers or, in general, of underwater magnetic sources that should ultimately support an automatic detection system which currently requires a human online monitoring or an offline signal processing. The research proposed here, by means of a windowing of the signals, uses Support Vector Machines for classification, as tool for the detection problem. Empirical results show the effectiveness of the method.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.