Geospatial Knowledge Management Approach to Automated Image Understanding

Information processing and knowledge representation are core components of knowledge management applications in C4ISR. Given the explosive growth of image data collection with associated text descriptors in these systems, Automated Image Understanding would advance the cause for automation urgently needed. Video imagery is emerging as a major source for intelligence information as well, especially in rapidly evolving operations. This transition from static to motion imagery is introducing substantial challenges related to the large amounts of data involved, and the corresponding application development requirements. Here, Milcord LLC and our partner intuVision propose to research, design, and build a feasibility prototype of an automated image understanding architecture using a geospatial knowledge management approach. Our approach features fully automated extraction of feature content from diverse sources, storing these features in an innovative integrated geospatial database, establishing the relationships between ancillary data taxonomy classes to build an ontology that defines, belief networks that uses the extracted source content in conjunction with the ontology for automated image understanding. Building our information processing and knowledge representation approach on previous US government funded R&D technologies developed by our team members will leverage these works and give our project a jump start in developing and operational prototype.