Image Fusion in Remote Sensing

Methods for fusing multispectral low-resolution remotely sensed images with a more highly resolved panchromatic image are described. The goal is to obtain a high-resolution multispectral image which combines the spectral characteristic of the low-resolution data with the spatial resolution of the panchromatic image. Unlike other applications, e.g. image fusion in military missions or computer-aided quality control, the main constraint in remote sensing is to preserve the spectral information for tasks like classification of ground cover. In the second part of the lecture a technique is demonstrated which avoids the requirement of an additional more highly resolved panchromatic image. The developed method is based on a sub-pixel misalignment of simultaneously acquired spectral bands and uses the displacement as additional information that effectively increases the sampling frequency. As an example, the depicted fusion algorithms are applied to three related multispectral images of a SPOT (Système Probatoire d’Observation de la Terre) satellite scene and the results are compared.