A Novel DP-TBD Algorithm for Tracking Slowly Maneuvering Targets Using ViSAR Image Sequences

摘要

We propose a novel algorithm, namely particle filter based dynamic programming track before detect (PF-DP-TBD), to track multiple slowly maneuvering targets using video Synthetic Aperture Radar (ViSAR) image sequences. In the proposed PF-DP-TBD algorithm, an independent particle filter (PF) is adopted around each candidate target. Each PF consists of a number of particles, and evolves according to the dynamic model of targets. A number of predicted states in each PF are selected as the predicted states at each time based on the posterior probabilities. The searching region during the dynamic programming (DP) operation is expanded adaptively. Experimental results show that the proposed algorithm can effectively improve the detection and tracking performance of dynamic programming track before detect (DP-TBD).