中文摘要：

从像素级、特征级和决策级遥感图像融合的角度,概述了遥感图像融合的研究现状处于瓶颈发展时期。提出遥感图像融合研究的三方面主要困境：缺乏统一的融合理论框架作指导;缺乏面向应用的融合算法设计;融合数据源的选择没有针对性,缺乏针对不同传感器数据和不同分辨率的融合研究。指出目前遥感图像融合研究在特征级和决策级融合,多角度融合,数据预处理精度,以及不确定性分析方面的研究不足。最后,提出了未来遥感图像融合的发展趋势和研究热点。

英文摘要：

The remotely sensed image fusion study was reviewed based on the pixel level,feature level,and decision level fusion.It was pointed out that the current status of remotely sensed image fusion is in the bottleneck period of the development.Three major aspects of dilemmas of remotely sensed image fusion were pointed out,which are the lack of general fusion theory as the guidance,the lack of application-orientation fusion algorithms designs,and the absence of fusion methods for the specific of data sources,especially the fusion algorithms for data of different sensors and different resolutions.Then the inadequacies of remotely sensed image fusion at feature level and decision level,multi-angle fusion,accuracy of pre-processing,as well as the analysis of uncertainty were pointed out.Finally the tendencies and prospects of remotely sensed image fusion were given.