Geospecific Night Imagery for Real-time Training Simulators

At present, display of nighttime lighting is primarily geotypical because of the cost to generate geospecific databases manually. However, the combination of two technologies can permit automatic extraction of light source data, including latitude, longitude, and elevation. These two technologies are aerial photogrammetry and pattern recognition. As part of a previous Phase II SBIR effort, DVC developed software to extract buildings, trees, and water features from 2D top-down imagery. In addition, software was developed to detect and remove time specific artifacts, such as shadows, vehicles, and both fixed wing aircraft and helicopters. This experience puts DVC in an excellent position to successfully meet the requirements of this SBIR. Diamond Visionics will provide useful solutions to address the challenges stated above based on our extensive background in the extraction of objects from 2D imagery. In Phase I, Diamond Visionics will develop a proof-of-concept PC-based software application capable of automatically extracting light source data from 2D imagery. The resulting data will be saved in a format generally utilized by both the GIS and training communities.