Preliminary evaluation of imaging sonars for high-security underwater assessment

This paper discusses preliminary evaluation of three acoustic systems capable of producing human-recognizable images of underwater scenes for use in high-consequence security system situations where final assessment of an event must be performed by a human operator. Tested systems included the BlueView ProViewer P450E-15, BlueView ProViewer P900E-20, and RESON SeaBat 7128, which were chosen for the good balance of resolution and range provided by their frequency of operation (450 kHz, 900 kHz, and 400 kHz, respectively) revealed during a prior study. Video clips were recorded from these systems while observing swimmers, open-circuit scuba divers, and closed-circuit rebreather divers in a marine harbor in St. Petersburg, FL. Clips were randomized and presented to 18 volunteer evaluators who were asked to identify which clips contained divers and which did not. The highest scoring system also had the highest operating frequency. However, the same high-scoring sonar in a different configuration scored similar to the system with the lowest frequency. This indicates that signal frequency, and therefore image resolution, is not a definite predictor of human assessment performance. After noticing many evaluators performed poorly with cluttered clips or clips with low acoustic target strength, results were recalculated to remove swimmers and surface swim routes such that only uncluttered, clearly-reflecting targets were included. Some sonar scores were moderately improved by this change, while others significantly decreased. This indicated that target clarity is also not a definite predictor of human assessment performance. Further analysis is recommended to reveal all major factors affecting assessment to maximize quality and eventually bring underwater security assessment to a high level.