Automated Change Detection and Classification System

This paper describes an Automated Change Detection and Classification (ACDC) system being developed by the Naval Research Laboratory (NRL) for the Mine Warfare group at the Naval Oceanographic Office (NAVOCEANO). ACDC autonomously detects features in sidescan imagery, classifies the features, searches through historical and dynamic databases of previously detected features and performs change detection (i.e., determines whether the feature is new or pre-existing, relative to earlier surveys). The system also clusters the detected features and performs “area matching” to minimize false detections. NRL is investigating other applications, such as augmented cognition systems, warning and alert systems, and autonomous “declutter” mechanisms for electronic displays, which could leverage one or more components of ACDC.