Automated Texture Synthesis from Multi-spectral Satellite Imagery
Agency / Branch:
DOD / NAVY
In Phase II, the goal is to develop software capable of producing and rendering an extremely high-fidelity synthesized environment from Landsat imagery and other sources, expanding upon the Phase I proof of concept implementation. The use of realistic synthesized terrain imagery and features will provide higher quality training, thereby increasing aircraft lifetimes by shifting training away from aircraft to simulators. This capability will also allow training to occur in areas for which imagery and other data may not be available or may be of insufficient resolution or quality. Additionally, by incorporating detailed and correlated geometric representations of other vector data sources (e.g. roads) with the synthesized imagery, the quality of the visual scene will match or even exceed that of areas for which high-resolution geo-specific imagery is available. Furthermore, by performing texture synthesis and geometry generation at run-time, the software can leverage advances in graphics processor units (GPUs), resulting in significantly higher fidelity scene generation compared to traditional approaches, without requiring significant off-line processing.
Small Business Information at Submission:
Research Institution Information:
DIAMOND VISIONICS LLC
400 Plaza Drive, Suite-A Vestal, NY 13850
Number of Employees:
OLD DOMINION UNIV.
ODU Research Foundation, VMASC
1030 University Boulevard
Suffolk, VA 23435 2345
Nonprofit college or university