提出了一种基于区域分割的图像融合方法,先将待融合的图像按空间特性分割成相似度不同的区域,然后根据具体应用目的对每个区域采用不同的融合规则.对多光谱与全色图像、多光谱与雷达图像两类遥感图像对进行了融合实验,实验结果表明,引入区域分割后的融合结果不但性能更优,而且可以很方便地控制不同图像源的成分对融合结果的贡献,满足特定应用的需求.

A region-based image fusion method was proposed.The source images firstly were split into different regions with various spatial characteristics and then performed the different fusion methods for each particular region independently.Experiments were implemented with two categories of remote sensing image fusion—multispectral image vs.panchromatic image and multispectral image vs.radar image.The results show that using the proposed method which could be not only obtained better quality fusion result,but also easily controlled the contribution of these components in the fused product and chosen the appropriate fusion method,which satisfied the user′s needs.