多源信息融合技术研究综述

A survey of research on multi-source information fusion technology

多源信息融合即基于多种(同类或异类)信息源,根据某个特定标准在空间或时间上进行组合,获得被测对象的一致性解释或者描述,并使得该信息系统具有更好的性能。从融合级别上来说,融合模型通常从数据、特征、决策三个层次上进行信息的融合处理。采用信息融合技术的系统结构一般可分为集中式融合、分布式融合和混合式融合架构。针对实际问题,根据信息源数据特征的差异,可单独采用不同层次的融合方法或组合某两个层次的递进融合方法,从而得到使系统性能较优的融合方案。

Multi-source information fusion technologies combine multiple homogeneous or heterogeneous information sources in space dimension or time dimension according to one specific standard,and obtain consistent interpretation or description of the measured object,then have better performance of the information system.According to the level of fusion,the fusion model usually integrates information from three levels-data level,feature level and decision level.The system architecture using information fusion technology can be generally divided into three kinds,such as centralized fusion,distributed fusion and hybrid fusion architecture.According to the actual problem with the different information source data characteristics,different level fusion methods could be separately adopted or some two levels of fusion methods combined,and better system performance could be obtained.