Title :
Track-centric metrics for track fusion systems
Author :
Canavan, Robert ; McCullough, Claire ; Farrell, William J., III
Author_Institution :
Dept. of Comput. Sci. & Eng., Univ. of Tennessee at Chattanooga, Chattanooga, TN, USA
Abstract :
In order to reduce the labor-intensive nature of evaluating quality of fused tracks in a multi-sensor, multi-platform environment, the metrics assessment system (MAS) was developed to provide heavily automated, real-time metric calculation and display. The MAS presents metrics in tables and graphs within the tactical display and provides overlays on the tactical display for focused assessment. Originally, MAS included standard kinematic metrics, but more recent work has generated additional metrics based on track-to-track versions of track purity, target effectiveness, and assignment accuracy, metrics derived from a confusion matrix analysis of the fused and known truth tracks, and receiver operating characteristic curves. Data mining tools were used to assess the effectiveness of candidate metrics for inclusion in the MAS. This paper describes the metrics, the evaluations, and the framework for metric assessment which this work has provided.
Keywords :
data mining; sensor fusion; assignment accuracy; data mining tools; metrics assessment system; receiver operating characteristic curves; standard kinematic metrics; tactical display; target effectiveness; track fusion systems; track purity; track-centric metrics; Computer displays; Computer science; Data mining; Kinematics; Performance analysis; Performance evaluation; Performance gain; Sea measurements; System testing; Target tracking; Tracking; data mining; metrics; performance evaluation; track fusion;
Conference_Titel :
Information Fusion, 2009. FUSION '09. 12th International Conference on
Conference_Location :
Seattle, WA
Print_ISBN :
978-0-9824-4380-4