Spatiotemporal trajectory models for metalevel target tracking

Abstract

Physical sensor-based target tracking is a classical problem that has been studied in great detail [1], [2]. This article presents metalevel tracking middleware algorithms to help human radar operators interpret tracks in order to detect and visualize suspicious spatiotemporal target trajectories. While state space models are ideal for target tracking, the main idea in this article is that stochastic context-free grammar (SCFG) models are also useful for modeling and interpreting trajectories.