PanSharpening

Pansharpening is a data fusion technique used to create a high-spatial-resolution multispectral image by combining a high-resolution panchromatic (grayscale) image with one or more lower-resolution multispectral images from the same scene. The goal is to preserve both the fine spatial detail of the pan image and the rich spectral information of the multispectral data.

Methods are broadly categorized into component substitution and multiresolution or transform-domain approaches. Component substitution methods, such as intensity-hue-saturation (IHS), Brovey transform, and principal components analysis (PCA), replace a spectral component with the pan image after appropriate scaling and histogram matching. Multiresolution methods—using wavelets, Laplacian pyramids, or similar decompositions—inject high-frequency spatial details into the multispectral bands while attempting to minimize spectral distortion.

Pansharpening requires well-registered data and compatible radiometric calibration between the pan and multispectral images. It is commonly applied to satellite sensors that provide a sharp panchromatic band alongside multispectral bands, such as Landsat, SPOT, Ikonos, and WorldView. The number of spectral bands, sensor characteristics, and processing parameters influence the trade-off between spatial resolution and spectral fidelity.

Quality assessment uses metrics that balance spatial and spectral performance, such as the QNR index, ERGAS, and spectral angle mapper (SAM). Evaluations compare sharpened outputs to reference data when available, or rely on no-reference indicators. Limitations include potential spectral distortion, color artifacts, and edge ringing, especially with aggressive sharpening or misregistration.

Pansharpening supports applications in land cover mapping, urban planning, agriculture, and environmental monitoring by providing high-detail imagery without sacrificing spectral information.