Fusion of Multiple Motion Information Sources

Our approach addresses the four Vs of big data and the four pillars, or principles, of ABI, and includes multi-INT data association and fusion, inclusive of objects and activities characterized from FMV streams.To address velocity and veracity associated with of multi-INT data sources we propose to develop a scalable graph-based analytic approach which will perform near real-time analysis of multi-INT data sources.Kinematic, contextual, and feature data can be updated in the graph as it is collected and processed.The graph accounts for the uncertainty of data without having to make hard data association decisions in the fusion process.Graphs can be extended to have fuzzy relationships between sets to account for uncertainties.Our graph-based fusion and analytics approach is extendable to more focused analyses commonly encountered by analysts at NASIC and NGA.The inclusion of 4d geo-spatial analytics will be used to combine pattern-of-life information with the results of graph searches in an intuitive way.To address the variety and volume of data we propose to architect a distributed storage and processing system that integrates graph technology.We will build on prior work that utilized the Hadoop ecosystem (HDFS, MapReduce, and Hbase).