Abstract:

To reduce the spatial and spectral distortions occurring in fusing remote sensing images, a novel pansharpening method using rolling guidance filter is proposed. The fusion model is designed with principal component analysis (PCA). In order to extract the small-scale detail information from a panchromatic image and large-scale structure information from the first component of a multispectral image, the scale-aware capability of a rolling guidance filter is employed. Then the detail information and structure information are combined. Through the inverse PCA transformation, the pansharpened result is obtained. The experiment is carried out on GeoEye-1 satellite data. The subjective analysis and objective evaluation show that the proposed method can greatly improve the clarity of the fused image while maintaining the spectral information, which is superior to some existing pansharpening methods.

Key words: Rolling guidance filter, Image fusion, Panchromatic image, Multispectral image