Computer-aided detection of rapid, overt, airborne, reconnaissance data with the capability of removing oceanic noises

Abstract

There have been three times more attacks to naval ships using sea mines than all other forms combined. Sea mines have always been viewed upon as underhanded and unchivalrous, yet they provide a weaker navy the capability to stall and damage a vastly superior navy. Utilizing unmanned sensors to detect sea mines is the goal of the navy for the future. Computer-aided detection (CAD) of sea mines is much faster and more consistent than a human operator, yet it is not currently being utilized by any of our mine countermeasure assets. Although there are many studies that have incorporated computer aided detection and classification algorithms with sonar imagery for mine warfare, few have used Light Detection and Ranging (LIDAR). During an amphibious assault scenario the ability to land assets quickly and mitigate risk is vital to the success. This thesis analyzes Rapid Overt Aerial Reconnaissance data from an Office of Naval Research experiment by Fort Walton Beach, FL. The CAD algorithm that was developed consistently detects sea mines in LIDAR data while having a manageable false alarm rate.