Visual Learning Systems (VLS) has been awarded a Phase I SBIR contract with the U.S. Army Topographic Engineering Center (TEC). With the award, VLS will develop technology that integrates its advanced machine learning techniques with rule-based classifiers for enhanced digital image analysis. The ensuing technology will fulfill a critical need within the Army to accelerate and improve its terrain analyses. Current approaches are cumbersome and require expertise in statistics, computer programming, and remote sensing; however, the VLS approach will extend its current system, Feature Analyst, which is easy-to-use, fast, and accurate. This approach will fuse information from multiple sources including multi-band imagery, terrain models, and GIS data layers to produce accurate geographic maps in support of the Army Warfighter.
In 2001 VLS released the Feature Analyst extension for ArcGIS to provide automated feature extraction and target recognition capability within the GIS workflow process. Feature Analyst is widely recognized in the GIS industry as the current standard in automated feature extraction and is now also available as an extension to Leica Geosystems’ ERDAS IMAGINE. During the research project, the underlying Feature Analyst architecture will be used to automatically generate and refine a rule base through an advanced machine learning technique called “theory refinement.” The system will leverage the feature extraction technology in Feature Analyst as well as the image processing capability of ERDAS IMAGINE Expert Classifier. Leica Geosystems will provide its expertise to VLS during this research contract.