图像融合是在多测度空间综合处理多源图像和图像序列的技术。图像融合分为: 像素层图像融合、特征层图像融合和符号图像层融合。本文讨论了图像融合各层次的融合算法以及红外图像、可见光图像、多谱图像、雷达图像等的融合问题。提出图像融合的结构模型, 并对图像融合的发展方向进行了展望。关　键　词: 　图像融合; 　结构模型; 　预测Abstract: 　M ultisensor image fusion is w idely recognized as a valuable mechanism for imp roving overall system performance in image2based app lication areas. The objective of image fusion is to combine information from multip le images of the same scene to ach ieve inferences that cannot be ach ieved w ith a single image or source. Th is paper surveys p ixel, feature, and symbol level image fusion methods and compares the characteristics of different fusion levels. A few image fusion algorithm s of each level are p resented, p roblem s about fusion of infrared image, V I, SAR, multispectral image are discussed. A n image fusion structure model w ith feedback information and some issues that should be solved in the future are p roposed in th is paper.KeyWords: 　image fusion; 　structure model; 　perspective