AURAL PROCESSED SIGNALS FOR MINE DETECTION AND CLASSIFICATION

THE DETECTION AND CLASSIFICATION OF UNDERWATER OBJECTS, SUCH AS MINES, BY MEANS OF ACOUSTIC SENSORS IS OF CONSIDERABLE PRACTICAL SIGNIFICANCE. UNFORTUNATELY, MANY RECENT ATTEMPS TO DEVELP SIGNAL PROCESSING TECHNIQUES TO ACCOMPLISH THIS AUTOMATICALLY HAVE ENJOYED ONLY LIMITED SUCCESS. RECENTLY, A NUMBER OF INVESTIGATORS SUGGESTED THAT THE CAPABILITY OF HUMAN LISTENERS CAN BE EXPLOITED TO FACILITATE THE DEVELOPMENT OF AUTOMATIC CLASSIFICATION ALGORITHMS. ONE OF THE MORE SUCCESSFUL TECHNIQUES WHICH EXAMINES HOW LISTENERS IDENTIFY PARTICULAR SETS OF SOUNDS INVOLVES OBTAINING AN ORTHORGONAL DECOMPOSITION OF A SET OF AUDITORY COMPARISONS USING MULTIDIMENSIONAL SCALING. THE PRIMARY OBJECTIVE OF THE PRESENT PROPOSAL IS TO EXPLORE THE POTENTIAL UTILITY OF HUMAN LISTENING CAPABILITIES FOR IDENTIFYING ACOUSTIC FEATURES USEFUL FOR MINE DETECTION AND CLASSIFICATION. MULTIDIMENSIONAL SCALING TECHNIQUES WILL BE APPLIED TO THE COMPARATIVE JUDGEMENT OF HUMAN LISTENERS. THE PROPOSED WORK WILL EXTEND EXISTING RESEARCH BY (1) INVESTIGATING SONAR RETURNS FROM SINGLE AND MULTIPLE TARGET OBJECTS OF DIFFERENT COMPOSITIONS AT DIFFERENT RANGES, (2) IDENTIFYING THE ACOUSTIC PARAMETERS BY WHICH HUMAN LISTENERS EVALUATE THESE SIGNLS, AND (3) EXPLORING THE FEASIBILITY OF DEVELOPING AN AUTOMATIC OR INTERACTIVE, SEMI-AUTOMATIC ALGORITHM TO ACCOMPLISH MINE DETECTION AND CLASSIFICATION.