Proceedings Article | 14 April 2008
KEYWORDS: Video, Video acceleration, Unmanned aerial vehicles, Human-machine interfaces, Video processing, Target detection, Computer architecture, Situational awareness sensors, Detection and tracking algorithms, Sensors
SET Corporation, under contract to the Air Force Research Laboratory, Sensors Directorate, is building a
Real-time Aerial Video Exploitation (RAVE) Station for Small Unmanned Aerial Vehicles (SUAVs). Users of
SUAVs have in general been underserved by the exploitation community because of the unique challenges of
operating in the SUAV environment. SUAVs are often used by small teams without the benefits of dedicated
personnel, equipment, and time for exploitation. Thus, effective exploitation tools for these users must have
sufficiently automated capabilities to keep demands on the team's labor low, with the ability to process video
and display results in real-time on commonly-found ruggedized laptops. The RAVE Station provides video
stabilization, mosaicking, moving target indicators (MTI), tracking, and target classification, and displays the
results in several different display modes. This paper focuses on features of the RAVE Station implementation
that make it efficient, low-cost, and easy to use. The software architecture is a pipeline model, allowing each
processing module to tap off the pipe, and to add new information back into the stream, keeping redundancy to
a minimum. The software architecture is also open, allowing new algorithms to be developed and plugged in.
Frame-to-frame registration is performed by a feature-tracking algorithm which employs RANSAC to discard
outlying matches. MTI is performed via a fast and robust three frame differencing algorithm. The user interface
and exploitation functions are simple, easy to learn and use. RAVE is a capable exploitation tool that meets the
needs of SUAV users despite their challenging environment.