基于字典学习的遥感影像超分辨率融合方法

Super-resolution fusion method for remote sensing image based on dictionary learning

鉴于多源遥感影像融合受现有分辨率的限制,结合稀疏表示理论,提出了一种基于字典学习的遥感影像超分辨率融合方法,可将多光谱影像的空间分辨率提升到全色影像空间分辨率的1倍或2倍。在遥感影像融合框架下,首先建立学习字典,利用冗余字典对影像稀疏表示,重构超分辨率;然后采用Gram-Schmidt(GS)光谱锐化法,融合得到超分辨率多光谱影像。利用QuickBird数据对提出的方法进行3个实验,结果都表明本文方法相对传统融合方法、传统超分辨率方法和其他字典学习方案具有一定优势,适用于遥感影像超分辨率融合,可为多源遥感影像融合的超分辨率问题提供1种可行的解决方案,而且对其他融合方法也有借鉴意义。

In consideration of the fact that multi - source resolution, the authors propose a super - resolution remote with sparse representation theory in this paper. The remote sensing image fusion is restricted by the existing sensing image fusion method based on dictionary learning spatial resolution of muhispectral images can be promoted to 1 or 2 times higher than the spatial resolution of panchromatic image. Under the framework of the method in remote sensing image fusion, a learning dictionary was established, the redundant dictionary on image sparse representation was used to conduct super - resolution reconstruction implementation. Then the Gram - Schmidt （GS） spectrum sharpening method was used as a fusion rule to obtain super resolution muhispectral image fusion. Three experiments were carried out using QuickBird data. The results show that the proposed method is suitable for remote sensing image super- resolution fusion with some advantages in comparison with traditional fusion method, traditional super- resolution method and the other dictionary solution for multi -source remote sensing image fusion, and has learning strategy. This paper provides a feasible referential significance for other fusion methods.