Validation of Automatic Ground Moving Target Indicator Exploitation Algorithms

ABSTRACT:The overall goal is to develop an evaluation systembased on real Mission dataof Ground Moving Target Indicator (GMTI) exploitation algorithms.Real-world Mission data is identified by analysts, who in the standard course of doing their job, fulfill Request for Information (RFI) Essential Elements of Information (EEI) responses.Our objectives are to provide the capability to evaluate a wide variety of GMTI exploitation algorithms; provide the capability to either evaluating the performance of a single algorithm or the evaluation of a collection of algorithms to determine the relative performance; provide a variety of event modification tools (such as, move events to other missions, align tracks to roads, and create synthetic events); provide statistical significance and evaluation of the algorithms performance, and use an automated machine learning approach of Relevance Vector Machine to provide a characterization of correct detections, missed detections, and false alerts.BENEFIT:TBDAnalysts are skeptical of automated tools that typically are tested on simulated data and lack the real world characteristics of vehicle motion, the interaction between vehicles, and the interaction of the sensor with the environment.Furthermore, special data collections are costly and are limited in scope and variety.Testing the algorithms with events contained in real world mission data, as well as performing statistically significant testing, will provide a realistic evaluation of the algorithms true performance under operational conditions.