Track-Before-Detect Based on Two-Stage Sampling Particle Filter

A modified track-before-detect based on two-stage sampling particle filter is proposed for detection and tracking of a target with low signal-to-noise ratio in an image sequence. According to each component´s impacting on measurement directly or not, the state vector is partitioned into two parts. Then a higher-dimensional sampling problem in TBD can be changed into two successive lower-dimensional sampling problems. This improves the efficiency of state sampling. Results on simulated data show a higher gain in performance of detection and tracking.