Smart System Classifies Data, Offers Options to Commanders

Enhanced versions of established tools assist imagery customers inundated with information.

As satellite data volume swells and virtual environments appear on more desktops, knowledgeable commanders are making split-second decisions by relying on their experience. Soon they will have the support of smart systems replete with subject matter expert material that describes choices and their consequences.

Geospatial information system and visualization users continue to search for better ways to understand and apply the image data they see to the situation in the field. The future of these technologies will be not only in providing the most detailed, up-to-date picture, but also in offering alternative actions based on the information at hand.

One company that specializes in the tools that allow people to wander virtually around a three-dimensional scene is making this next leap. This summer, ERDAS Incorporated, Atlanta, will release a software package that uses decision-making rules to teach computers to extract information, compare different sources of data, then offer next-step alternatives to users.

“Most systems are still manually intensive,” Kurt E. Schwoppe, manager of federal systems, ERDAS, says. “Viewers have to use the mouse to click on several sections of a road, for example. Right now, we’re working toward more automation. This is the state of the art that we’re trying to push.”

Work on the company’s Expert Classifier began three years ago with initial funding through the National Aeronautics and Space Administration small business innovation research program.

Inference engines are the core of the decision-making process, but the technology available at the beginning of the project was not effective at reviewing large amounts of raster or vector geospatial information system (GIS) information, Schwoppe explains. As a result, the firm developed its own inference engine, one that can process the voluminous quantity of GIS and other information that is available to and demanded by imagery customers.

The growing number of operating satellites increases the amount of available images, and their enhanced resolution results in larger amounts of data per image. The combination of readily available, high-resolution imagery and changing Defense Department missions is driving the expansion of this high-powered technology. It is precisely the velocity at which imagery information is pouring in that creates the need for interpretation speed. Improved technologies are bringing these pictures into the hands of many more people, including adversaries, at a brisk pace.

“You have to be faster at getting the information to U.S. forces if you want to lead—faster than the bad guys are able to get it—so you can make faster decisions. That’s why this capability is important,” Schwoppe offers. “The military needs real-time feed of imagery. Pictures of roads that are even hours old are already obsolete,” he adds.

Acquiring the images is only one part of the equation. To make this data valuable, a system had to be developed that could extract the items a commander needs. Add weather information to the geographic data, and a commander has the total picture. This data must then immediately be put into a user-friendly format so that information can be rapidly pulled and reviewed.

One application Schwoppe offers as an example is terrain analysis, used when planning military operations. When the U.S. Army needs to cross a specific territory, especially an area of denied access, a mixture of different types of imagery, extraction tools and an expert system can determine the stem diameter of the trunks and branches of the trees as well as the distance between trees. Additional sensor information can provide the tree type—for example, oak—helping tank commanders determine if the terrain will be easy or difficult to travel, he explains.

The Expert Classifier compares data from different sources and provides various options based on a set of rules. Feedback loops allow for additional alternatives. It is a suite of capabilities dedicated to understanding sensor systems and data. “We are trying to capture the knowledge of the scientist and put it into the system,” Schwoppe reveals.

“If you can get this information to the warfighter quickly, it is useful. Otherwise, the war could be over by the time you have the information you need. It’s the difference between the Vietnam and Desert Storm wars,” he adds.

The Expert Classifier is an element of the ERDAS IMAGINE product line, a suite of geographic imaging software that offers a range of functionality from data visualization to standard data manipulation algorithms to advanced image classification.

The company’s strengths lie in developing software that allows users to overlay images and compare multiple data sets then identify changes in an area, according to Ian S. Anderson, IMAGINE product manager, ERDAS. The military can detect changes in an area, highlight and digitize these changes, then store them for future reference. Imagery can be displayed at a scale determined by the user. By employing multiple windows that can be opened and simultaneously linked by geography, data sets can be compared.

Geographic precision is a key element in using GIS and visualization, Anderson explains. One IMAGINE software product introduced last year, OrthoBASE, allows users to create planimetrically accurate orthorectified images for seamless orthomosaics. “After all, the world is not a flat surface. Since the world is not flat, calculations must be exact and not taking into account the curvature, even by a little bit, could lead to inaccurate missile delivery,” he offers.

IMAGINE VirtualGIS is the ERDAS contender in the three-dimensional imagery visualization arena. It allows users to extend the geographic imaging functions in IMAGINE and experience real-time navigation through a large database of image, map and attribute data.

Visual navigation in VirtualGIS includes mouse-operated three-dimensional pan, zoom and flight control; interactive pivoting around a defined target point; user-defined flight paths; and head-up flight displays showing pitch and azimuth. Topographic surface parameters include the ability to change height exaggeration, separate layers of data in the z-axis direction, control sun shading, display multiple-resolution digital elevation model (DEM) surface layers, and include nontopographic DEM surface layers for dynamic visualization.

Early versions of the tool featured three-dimensional raster and vector draping capabilities. The raster drape allowed the coordination and simultaneous viewing of satellite imagery, photographs taken from aircraft, scanned maps and thematic images. The vector drape capability enables users to include the ESRI Incorporated software product ARC/INFO coverages or annotations over a surface and create buildings by extending this information in the z-axis direction.

Military units employ VirtualGIS in training exercises. The draping of geographic information layers over terrain data allows soldiers to plan missions or familiarize troops with inaccessible or hostile terrain. During after-action reviews of exercises, the tool’s video playback feature allows participants to critique performance and discuss “what-if” scenarios.

Late last year, the company introduced an updated release of the product, VirtualGIS 8.4, a visualization and analysis tool set that operates on both UNIX and Windows platforms. This latest version adds intervisibility calculations that offer a three-dimensional mesh of what would be seen, or not seen, from a given position. While previous versions of the tool allowed users to drape raster and vector information then extract and pretend to have these changes, they could not actually be incorporated into the scene, Anderson explains. These changes can now be added to the three-dimensional scene.

New features added to this latest edition include interactive three-dimensional models, virtual reality modeling language 2.0 support for creating three-dimensional web content, real-time anaglyph stereo and the ability to create virtual worlds of unprecedented size, company officials say.

IMAGINE version 8.3 or 8.3.1 is required to take advantage of version 8.4. The company will put version 8.4 updates on their web site, so clients will be able to update their tools by downloading them as they are released.

Future developments at ERDAS are likely to involve two areas: mobility and object software.

Currently, GIS information processing takes place at base camps, and these data are given to soldiers to take with them, Anderson explains. As the computer becomes smaller and lighter, troops will be able to take the processing capabilities with them and conduct analysis of raw information on the fly. These advances will begin taking place in approximately two years, he predicts and adds that the United Kingdom is already working on a palmtop version. Global positioning system (GPS) technology is a key part of development. “This is real analyzing of data, not just processing of images,” Anderson offers. Portable capabilities including a combination of GPS and GIS will protect ground troops from wandering into enemy territory, he explains.

Schwoppe believes that the speed at which technology is changing combined with the decreasing costs of hardware and software as well as their increased capabilities will continue to make commercial off-the-shelf products a viable option for the military. But because different military commands have some specialized requirements, companies must continue to develop and provide systems that consist of components that can be used separately or together. Interoperability with products from other vendors is also key.

“Offerings must be very customizable because that’s what the customers want. Also, customers want, and ERDAS is focused on providing, pieces that can be quickly pulled into systems,” Schwoppe says.