多分类器融合在水中目标识别中的应用

为了提高对水中目标的识别能力,研究使用DS证据理论对多分类器进行融合,论述以往确定BPA方法的优缺点,并在此基础上提出一种新的利用分类器性能和输出信任度来确定BPA的多分类器融合方法,实验证明这种方法在对水中目标进行识别问题中的优越性,为进一步研究分类器融合在目标识别中的应用提供参考.

Application of a Multi - classifier Fusion Method in Aquatic Target Recognition

In order to improve the recognition ability of aquatic target, this paper researches on fusing multi - classifier based on D-S evidential theory, alter analyzes the characteristic of old ways determining the BPA, this paper proposes a new fusion method utilize both classifier ability and output belief to determining the BPA . Finally, the classification experiments were performed and the results indicated that this algorithm has better classification ability than old ways. This paper provides some references for further research on classifiers fusion in target recognition.