基于Dempster-Shafer证据推理的水中目标多特征融合识别技术  

Water Target Multi-feature Fusion Recognition Technology Based on Dempster-Shafer Evidence Reasoning

作　　者：吴亮[1] 彭圆 闫祎 

机构地区：[1]海军驻大连地区军事代表室,辽宁大连116021 [2]大连测控技术研究所,辽宁大连116013

出　　处：《计算机测量与控制》2013年第9期2503-2505,共3页Computer Measurement &Control

摘　　要：多特征融合是将各种信息以某种优化准则组合起来,产生对观测目标的一致性解释和描述,从而形成比单一信息源更精取、更完整的估计和判决;文章重点研究了Dempster-Shafer证据理论,讨论了它的可应用性,在此基础上把决策融合策略与模糊自适应共振(FART)神经网络、所获取的特征知识相结合,实现了对三类目标的分类识别,实验表明通过决策融合后可使识别率比原来提高2%左右,提高了识别的可靠性,显示了该方法在实现水中目标识别上的重要应用前景。Many features fusion is all kinds of information in some rule of optimization combined, the consistency of the description and interpretation is given, so as to form more than a single source of pure take, more complete estimation and judgment. This article focuses on Dempster--Shafer evidence theory, discusses the application of it, and based on this, the decision fusion strategy and fuzzy adaptive reso- nance （FART） neural network, and the characteristics of combining for knowledge, realize the goal of three kinds of classification, the ex- periment that through decision fusion can make the recognition rate than the original increased 2~ or so, improve the reliability of the recog- nition, shows that the method in the realization of the target recognition of important on the application prospect.

关 键 词：Dempster--Shafer 融合识别 神经网络 水中目标 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]