2 February 2001
High-Definition TV Emerges As New Reconnaissance Tool
Source: Aviation Week &
New York; November 6, 2000
David a. Fulghum
Full Text: Copyright 2000 The McGraw-Hill Companies, Inc.
A new reconnaissance discipline--based on motion--is gaining visibility and some popularity in the intelligence-gathering community.
Movement intelligence (movint) is taking its place beside analysis of communications (comint), imagery (imint) and electronic signals (sigint) as a critical analytical instrument for plotting the actions and intentions of foes.
Of broad potential usefulness in generating movint is ``motion imagery''--the analysis and manipulation of high-definition digital television (HDTV) images. The technology has been developed over the last two years by researchers from the Naval Research Laboratory (NRL) and SGI Federal at the former's motion-imagery laboratory in Washington.
Researchers are producing high-quality images that can be exploited because of four other modern technologies--HDTV that can record images without blurring, wide-band data links for moving the images, compact storage packs for archiving large amounts of imagery, and advanced algorithms and computers that can quickly scan the images for high-value targets or activity of interest.
Researchers have used HDTV cameras to record images of military targets from the air, shipped and stored the images in real time, cleaned the images of weather artifacts such as haze, magnified them by as much as 40 times and enhanced the images for improved clarity.
The surveillance system would likely fit between satellites and short-endurance reconnaissance aircraft, said Henry Dardy, chief scientist for NRL's Center for Computational Science. ``You [could] deploy it on a fleet of planes [like high-altitude, long-endurance UAVs] that fly from 50,000-60,000 ft. and [thus] can really survey the battlefield and [immediately] send the video footage back.'' However, researchers believe the system also could be adapted both to satellites and tactical reconnaissance aircraft.
A number of associated algorithms also allow development of some unique forms of surveillance. The system has a ``stake-out capability'' that allows an operator to isolate a small area, such as the door of a building or the entrance to a cave. The area is then automatically monitored while the operator searches other parts of the battlefield. The operator is alerted to return to the critical site when, for example, a person walks through the door or a mobile missile launcher emerges from the cave.
``Two years ago, we couldn't do this,'' Dardy said. ``We didn't have a way to slosh through tera-bytes of imagery. Now we can offer the battlefield commander tele-presence with high-end imagery.'' The ultimate goal is to replace film, he said, in a system that could be available to the Pentagon and intelligence and research agencies as early as 2004.
Movint has already become a military staple. It is being produced by the E-8 Joint-STARS, which first saw combat service in 1991 when a prototype was rushed into the Persian Gulf area. It was the first to spot the Iraqi armored attack on the Saudi Arabian town of Khafji, an attack made under the cover of darkness. The aircraft's 24-ft.-long radar antenna can locate and track the movement of ground vehicles and slow-moving helicopters at about 200-mi. range. An improved version of the system is under development that would allow operators to pick up much smaller targets and fast-moving aircraft including small, stealthy cruise missiles at ranges of 200-300 naut. mi. Another moving target indicator (MTI) system is flying on the latest version of the U-2.
Smaller variants of the MTI radar are also in development for the long-endurance Predator and Global Hawk unmanned reconnaissance aircraft. Moreover, momentum is growing to put a movint system into space on board a constellation of 24 inexpensive (less than $100 million each) low-Earth orbit satellites.
Researchers and military planners want to push movint into additional parts of the electro-magnetic spectrum, in particular visual imagery like that from electro-optical and infrared sensors. The problem is that standard television and infrared cameras, when frozen on a single frame, produce blurry images. Freezing a frame of video destroys both the crispness of the standard moving image and the eye's ability to patch together a sharp image from several frames. Yet this single-frame format is necessary for analysts to scour an image for intelligence clues.
NRL and SGI researchers think they have licked both problems with live, high-definition cameras that can scan in real time while frames of interest can be isolated and exploited for additional information. These recent technological advances are allowing the use of movement intelligence to spill out of the radar frequencies. With an acuity that is at least four times that of conventional television, digital HDTV images can be exploited easily through computer manipulation.
A key decision by developers was to present the imagery in a progressive imagery format rather than interlaced video. Interlacing is the system used by television today where alternate lines show the next frame of the video. Progressive television uses a complete, single-frame image and then moves to the next complete frame. By using HDTV as the baseline, researchers offer a new starting point for the use of motion imagery in reconnaissance and surveillance.
They intend to eliminate blur and jitter in motion imagery (no longer called video) by using a digital camera that gathers data with more than 1 million pixels at a rate of 60 frames per sec. The system also employs bilinear and bicubic filtering of pictures that become blocky once they are greatly blown up. Bilinear filtering manipulates the nearest four pixels around an element in the picture to sharpen definition. That image could be the outline of an aircraft, ship or building. It is effective in improving high-speed pictures. When less motion is involved, bicubic filtering is used to alter the nearest 16 pixels.
``As you are blowing up the image, it [the filtering algorithm] is looking at the nearest pixels and doing interpolation'' to better define lines and other detail, said Alan Dare, SGI Federal's geospatial imagery solutions manager. ``The analyst can pick the various filter combinations to give him the best results based on the way the footage is gathered, the clutter and the speed of the footage itself.''
To quickly move the improved motion imagery, NRL and SGI researchers are working on a secure communications network that would eventually be linked into a Vision Global Grid that would allow planners or war fighters to pull down data about any place in the world. To keep costs down, parts of the data would be available to the commercial market over commercial communications circuits. Yet classified users of the grid would have added protection and a secure network. As the system is developed, researchers are building in the hooks that will let them add end-to-end authentication for users of the classified portion of the system.
The ultimate challenge of building the system will be the ``need to give [commanders and intelligence analysts] the data in a way they can use quickly and effectively,'' Dardy said.
The sensors and computer network system--offering moving gigabit stream video--will first operate as an advanced technology demonstration being funded by the Defense Advanced Research Projects Agency's (Darpa's) next-generation Internet research program. Other participants on the Washington net are NASA, the National Security Agency, the Defense Information Systems Agency and the Defense Intelligence Agency.
Also propelling the system is the development of large bandwidth communications. While the demands of computing and imagery collection are stressing today's computers, ``the growth in network capacity is stripping everyone's expectations,'' Dardy said. By the end of the demonstration, he predicts that improved technology will exceed the current system's ability to transmit at 1.5 gigabits per sec. over a single line and 17 gigabits per sec. aggregate transmission using commercially available computers and data links. ``We can grow as fast as the industry is growing,'' he said. Currently, NRL and SGI are using a Silicon Graphics Onyx 2 computer.
Once feasibility of the communications and archiving network is established, researchers say data from aircraft, spacecraft and ground sensors can be poured into the system. Overlays of multispectral data will allow analysts to deal with knotty problems like hidden, buried and camouflaged targets. Moreover, SGI researchers say they can change the format of the system so allies with legacy computers and sensors can tap into the system and retrieve data in a useful form.