Fusion of multispectral and panchromatic images based on a novel inter-band structure model

Image fusion is one of the most important image processing methods in the field of remote sensing. Multispectral (MS) or hyperspectral (HS) images are often fused with panchromatic (PAN) images to enhance their spatial resolution while preserving the spectral information, which will lead to a better interpretation in subsequent applications. In order to achieve this goal, an improved method of image fusion, which is based on the amélioration de la résolution spatiale par injection de structures (ARSIS) Concept, is proposed in this paper. In the new method, the degree of diversity between each pixel and its surroundings is measured utilizing Center-Surround Spectral Angle (CSSA) model, and the more different a pixel is, the more detailed information is injected into the MS image. In order to verify the advantages of our proposed fusion method, experiments are processed on two data sets. Several assessment criteria are used to demonstrate the effectiveness of our proposed algorithm. Moreover, from the view of application, accuracy of classification on fused images is also evaluated to further show the superiority of our method.