Abstract:
Information fusion techniques, which can help to effectively reduce or eliminate the measuring uncertainty of distributed sensors′ signal and fuse more comprehensive original vessel radiated signals, have been widely used in vari-ous military and civilian fields, and have attracted more concerns in the world. In this paper, the existing most accepted function models of the fusion systems are summarized, and a three-level (data-feature-decision) underwater automatic target recognition (ATR) system model is proposed. Subsequently, several commonly used fusion algorithms based on the probability theory, the data classification theory, and the artificial intelligence theory are presented, and their advan-tages, disadvantages and application constraints are analyzed. Moreover, the development trend of the underwater target fusion recognition system based on multi-sensor system, the existing problems, and the solutions to these problems are all discussed.