Computer aided detection/computer aided classification and data fusion algorithms for automated detection and classification of underwater mines

Abstract

Raytheon has successfully developed a computer-aided detection/computer aided classification (CAD/CAC) algorithm to process the sidescan sonar outputs of both the AN/AQS20 helicopter-towed minehunting system and Woods Hole Oceanographic Institute's (WHOI) Remote Environmental Monitoring UnitS (REMUS) unmanned underwater vehicle. These systems employ high frequency acoustic imaging sonars to detect, classify, and localize minelike objects on the ocean bottom. The algorithm was initially demonstrated at the Coastal System Station (CSS) underwater range in Panama City, Florida, and then applied to REMUS sonar imagery taken in the Very Shallow Water (VSW) environment off the coast of San Diego, California. A data fusion technique for combining the outputs of three different CAD/CAC algorithms was subsequently developed and applied to a set of REMUS data. The fusion demonstrated a 4:1 reduction in false alarms relative to any single CAD/CAC algorithm. This paper gives overviews of the AN/AQS30 and the REMUS systems, describes the Raytheon CAD/CAC and Data Fusion algorithms, and gives sample results from processing of the sea test data.