High-Speed Image Recognition Control System

摘要

Terminal guidance of weapons immediately prior to impact is a demanding control problem. While Global Positioning System (GPS) guidance can get a weapon close to a target, a greater level of accuracy is often needed to account for such uncontrollable factors as the movement of a target. The inclusion of an alternate terminal guidance ability therefore will increase the chances of the missile reaching its target. This final report details results of a Small Business Innovation Research (SBIR) Phase I project to develop a novel, extremely high speed image-recognition control system that seamlessly links an image sensor input, target identification and clutter discrimination processes, and control outputs. The missile guidance architecture evaluated in this project is based on a novel mathematical methodology that eliminates many faults or limits found in existing control architectures. Mathematical modeling of the system demonstrates that it can track both simple and complex images on backgrounds that range from plain to fairly cluttered. Additional refinements are needed to improve the system's ability to discriminate between target and decoy images.