Abstract:

Based on the application of the D-S evidence theory in the single sensor emitter recognition, a method suitable to single sensor emitter recognition based on combining classifier fusion (CCF) is put forward. It uses two kinds of knowledge, one is prior knowledge which is characterized by the classification ability of combining classifier to the samples and the other is current knowledge which is characterized by the classification ability of combining classifier to the unknown patterns. The correctness of current knowledge is improved at the classifier fusion input site based on the real time updating of the prior knowledge to the current knowledge. The correct information of the classifier at the output site is extracted validly by the combining arithmetic and the single sensor emitter recognition accuracy is improved finally. The simulation results show the feasibility and validity of the method which is referential for the engineering application.