[关键词]
[摘要]
本文用信息论的观点,给出了在设计单个分类器时使系统性能得以提高的条件,基于群组决策理论和模糊积分方法提出了一种多证据条件下进行信息融合的方法.该方法被用于被动声纳目标分类,数值模拟表明了该方法的有效性.
[Key word]
[Abstract]
The conditions for designing an individual classifier, which could increase the performance of combined classification system, are obtained in the opinion of information theory in this paper. The information fusion of combined system in conditions of multiple evidences is implemented by using fuzzy integral. This approach is applied in sonar target classification. Simulation results show that it is efficient.
[中图分类号]
[基金项目]
本文研究得到船舶工业科技基金资助.