Advances in Prediction Performance Modeling of Multitarget Tracking Systems

The document discusses target detection performance in overhead imagery used for detection and tracking applications. The imagery consists of sequences of frames from a fixed vantage point, described mathematically to include background scenes, target positions, sensor platform jitter, and additive noise. The document addresses background clutter modeling, considering realistic scenes with non-Gaussian noise. The performance prediction model is updated with clutter noise statistics that are derived from a realistic scene and background suppression technique. The background models are used with previously developed performance prediction models that predict the detection and tracking performance of an overhead imaging system. This will allow operators to properly task systems with a suitable geometry and sensor exposure to meet desired detection performance. The predicted detection performance is verified by Monte Carlo based simulations.