Autonomous Multi-Sensor Registration Toolkit

Future strike missions will need to utilize intelligence, surveillance, and reconnaissance and other off-board data assets to improve the geo-location, detection and identification of targets. To share and fuse data from multiple sources, it is necessaryfirst to reference the different sources of information with respect to a common coordinate system, which is the process of data registration. This proposal addresses the multi-sensor data registration problem for both image and report level sensor data.The key challenge for multi-sensor registration is the automatic identification, extraction, and association of salient features across the different sensor data sources to determine tie-points for bundle adjustment. To address this challenge, thisproposal presents a mathematically-based, unifying framework for data registration that allows for the plug and play of multiple registration algorithms, each focused on the solution for its portion of the multi-sensor data registration problem space.Phase I will provide an initial proof of concept of the data registration framework using a limited multi-source data set. Phase II will integrate a complete suite of algorithms into this unifying framework and demonstrate over a wide range of datasources. The technology developed under this program has both military and commercial applications. Military applications include platform and/or ground station geo-registration services. These geo-registration services can be integrated with existing sensorexploitation tools and provide enhanced performance for various processes like targeting, tracking, fusion, and change detection. Commercial applications include geo-registration services for Geospatial Information Systems (GIS) applications andgeographic image analysis tools. The product of this effort can be integrated as a plug-in to commercial applications such as ERDAS IMAGINE and ESRI's ArcView to enable the automatic alignment of different data sources.