An information fusion algorithm for data association in multitarget tracking

We employ the technique of uncertain information processing to solve problems of multitarget tracking. We consider the data association problem as a fuzzy partition. Dempster-Shafer theory is used to evaluate the plausibilities of the association events. Using the plausibilities, a fuzzy partition is performed. The grade of membership is then used as the weight of data association. A radar and a passive sonar tracking of two crossing targets are studied through computer simulation respectively. The results show that the Dempster-Shafer theory based approach has an excellent ability to track multiple targets and has less complexity of the computation than the JPDA's.