Aircraft target recognition based on recursive inference of fuzzy discrete DBNs

Nowadays dynamic Bayesian network (DBN) has been well known for its marvelous capabilities in modeling and analyzing a wide range of sequential systems. The notable advantage of DBN lies in it provides a good solution to the inference problems caused by uncertainty and complexity. Usually the main reason for uncertainty is randomicity; however, sometimes it could also be caused by vagueness. Since fuzzy logic is an effective tool to solve the latter problem, this paper presents the idea of merging fuzzy classification with discrete DBN, which improves the representing as well as reasoning abilities of discrete DBN and enables it to deal with the inference problems concerning continuous observational data. Finally, the validity of fuzzy discrete DBNs combined with a recursive inference algorithm has been demonstrated by an aircraft target recognition example.