New efficient target tracking based upon hidden Markov model and probabilistic data association

This paper deals with automatic detection and tracking using hidden Markov model and probabilistic data association in order to operate in a densely cluttered environment. After a theoretical description of the algorithm, Monte-Carlo performance comparisons with known methods like NNAF and PDAF are provided, in the case of sonar processing. Improvements are clearly shown in terms of detection and track splitting, for an increase of computation requirement less than 5. At sea signals analysis confirms this result.