Volume 11, Number 1 • Spring 2003 • Small Business/SBIR

Changing the Rules in the GIS Community

A state-of-the-art, machine-learning, feature-extraction software system developed for the Geographic Information Systems (GIS) industry under the NASA Small Business Innovation Research (SBIR) program at Stennis Space Center is proving to be the key to a paradigm shift for extracting geospatial information from digital imagery. Developed by Visual Learning Systems, Inc. (VLS) of Missoula, MT, the Feature Analyst is a software extension for leading commercial-off-the-shelf (COTS) GIS and image-processing software, such as ESRI’s ArcGIS™ and ERDAS Imagine®, which automates the geospatial feature-extraction process from digital imagery. Widely regarded as the first viable Automated Target Recognition/Automated Feature Extraction (ATR/AFE) system, Feature Analyst reduces labor costs, is more accurate than existing image-processing techniques and hand digitization, and is simple enough for a technician to use.

Geospatial features, such as streets, buildings, landmarks and vegetation, are used in the GIS industry to produce maps and perform spatial analyses for urban planning, transportation analyses, defense, telecommunications and many other applications. The Feature Analyst technology speeds up this feature-extraction process using VLS’s proprietary machine-learning algorithms. This innovative technique cuts labor costs on digitizing projects by 80 percent or more while fitting into the existing commercial GIS software workflow process. As an ATR/AFE toolkit, the Feature Analyst will increase the efficiency of a GIS database by a factor of 50 to 200 times over traditional heads-up digitizing methods, with an accuracy that surpasses expert heads-up digitizing.

Feature Analyst was developed to assist NASA’s critical need to accelerate and automate the identification and classification of features in digital satellite imagery to support its Earth Science Enterprise mission. Commercial applications of NASA’s Earth science data depend on the development of new Information Technology tools that accelerate the analysis of remotely sensed images to support geospatial data models for land-cover and land-use change.

Primary markets for the Feature Analyst software include the US Department of Defense and Homeland Security Operations, NASA and the National Imagery and Mapping Agency. These agencies require timely, accurate and relevant GIS data to support intelligence and emergency planning operations. Other markets include civil government, transportation, forestry, environmental and agricultural.

VLS has conducted market research that shows, based on the proposed Feature Analyst extension for ESRI’s Arc 8 technology alone, the market for the proposed Feature Analyst software will garner between $25 million and $50 million annually in 10 years.

Visual Learning Systems and NIMA recently conducted a detailed quantitative analysis of the Feature Analyst based on written test procedures provided in the NIMA Automated Feature-Extraction Algorithm-Validation Program.

The performance evaluation consisted of a series of tests in which the cartographers extracted a variety of geospatial features using both the Feature Analyst and manual methods. Objective data were collected on the timing and relative accuracy of the extraction of vegetation and double-line drains. The accuracy of the Feature Analyst software was superior to hand digitizing in all instances. Based on these results, VLS has reinvested $150,000 for continued development of the product and entered into a business partnership with the Environmental Systems Research Institute (ESRI) of Redlands, CA, the world’s leading developer of GIS software products. VLS plans to make several future enhancements to the software, including a change-detection module for detecting changes over time, georegistration of old vector layers to new imagery in GIS databases, and incorporation of three-dimensional information into the feature-extraction process.

“The SBIR Program is a highly competitive program that is widely recognized as selecting the best-in-breed of innovative small businesses poised to solve federal government technology problems and develop new commercial products around those concepts. Visual Learning Systems, Inc. is a performance-driven organization that thrives on competition, innovation and challenges,” said Stuart Blundell, chief operating officer for Visual Learning Systems, Inc. “The SBIR Program was a natural outlet for VLS to pursue federal grant opportunities to drive the Feature Analyst technology into the commercial GIS market.” Q

For more information about Feature Analyst and a free trial version of the software, go to www.featureanalyst.com or contact Visual Learning Systems, Inc., T: 406/829-1384, sales@vls-inc.com. For more information about SBIR opportunities at Stennis Space Center, contact Bill Graham, T: 228/688-1889. Please mention you read about it in Innovation.

NASA Official: Jonathan Root • Web Design: Printing & Design Office, NASA Headquarters • Credits