针对证据冲突状态的图像融合技术  

Image fusion technology for conflicting information condition

作　　者：廖家平[1] 陶靖琦[1] 赵熙临[1] 

机构地区：[1]湖北工业大学电气与电子工程学院,湖北武汉430068

出　　处：《华中科技大学学报（自然科学版）》2012年第3期5-8,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61072130);湖北省科技厅国际合作项目(2010BFA007);湖北省教育厅科研基金资助项目(D20111410)

摘　　要：为解决D-S(Dempster-Shafer)证据理论在较大证据冲突下的结果失效问题,根据最大信息熵原理,提出一种基于马尔可夫过程的证据冲突处理方法.当证据间表现出冲突特性时,通过马尔可夫过程平稳分布获得冲突证据的一致性输出,再与后续证据进一步融合,获取客观科学的融合结果.红外序列图像融合仿真表明:提出的融合方法的收敛性与融合结果均优于经典D-S证据理论,对消除证据融合过程中的主观因素起到一定的作用.To solve problem of using Dempster-Shafer evidence theory to combine high conflicting information,the maximum entropy principle was introduced and a treatment based on Markov process was constructed to deal this problem.The consistency judgment of conflicting evidences was obtained from stationary distribution of Markov process,then by using Dempster rules remaining evidences were combined,and finally the object judgment was resulted.The performance between this treatment and Dempster-Shafer theory was shown,more effectively and better convergence than classical D-S theory,and the subjective factors which in the process of fusing were decreased in some extent.

关 键 词：图像融合技术 信息融合 D-S(Dempster-Shafer)证据理论 马尔可夫过程 最大信息熵原理 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]