A novel Multi-Bernoulli filter for joint target detection and tracking

Aiming at the detecting and tracking of the moving dim targets from image observations with low signal-to-noise ratio(SNR), this paper puts forward a new track-before-detect algorithm based on Gaussian particle filter implementation of Multi-Bernoulli filter (GPF-MB-TBD). GPF-MB-TBD can greatly reduce the requirement for storing, and improve the performance of tracking than the sequential Monte Carlo implementation of Multi-Bernoulli filter(SMC-MB-TBD). Simulation results show that this novel algorithm is suitable for detection and tracking multiple targets from image observations.