Target Recognition Based on Fuzzy Comprehensive Evaluation Using Improved Weight

LUO Lei,LI Yue-hua,LUAN Ying-hong,WEI Yong-ping
2009-01-01
Abstract:The traditional target recognition methods of millimeter wave(MMW)detector which are all based on template matching lead to high false alarm rate.The membership function of metal targets is determined based on the target's feature components by using fuzzy statistics and the evaluation matrix of fuzzy comprehensive evaluation is built.A new method of determining the weight of each feature components is proposed according to the two factors which are the recognition effect of single feature component and the expert scoring weight.According to the results of fuzzy comprehensive evaluation and maximum membership principle,the classifier is built by using training sets with the consideration of risk loss of the decision system.The recognition rates of the metal target samples under three different conditions are 80%,83% and 81% respectively.With the classifie,the metal target recognition rate obtained by high tower experiment and scale down experiment is 81% in general.
What problem does this paper attempt to address?