Real-time performance of fusion algorithms for computer aided detection and classification of bottom mines in the littoral environment

The fusion of multiple computer aided detection/computer aided classification (CAD/CAC) algorithms has been shown to be effective in reducing the false alarm rate associated with the automated classification of bottom mine-like objects when applied to side-scan sonar images taken in the littoral environment. Real-time operation of the CAD/CAC fusion algorithms from Raytheon, Lockheed Martin, and NSWC Coastal Systems Station (CSS) on board an unmanned underwater vehicle has recently been successfully demonstrated as part of a littoral test sponsored by the Office of Naval Research (ONR) in 2002. Test results proved that the fusion reliably classified bottom mine-like objects while significantly reducing the false alarm rate relative to that of a single CAD/CAC algorithm. This paper discusses the hardware and software architecture for the real-time implementation of the CAD/CAC algorithms, and presents the real-time performance results obtained during the experiment. Additional post processing performance results are also discussed for alternate fusion approaches, and the overall performance benefit through a significant reduction of false alarms at high correct classification probabilities is quantified.