Recently, the development of sensors has made it possible to acquire satellite images with various spatial and spectral resolution. Since these satellite images have different characteristics from existing high spatial resolution satellite images, there is a limit to apply existing pansharpening methods.
In this study, a sharpening method was developed in order to minimize spectral distortion and improve the spatial resolution of satellite images with various spatial and spectral resolution. When a panchromatic image was not provided during sharpening, a synthetic panchromatic image was produced using a selected band scheme and synthesized band scheme.
The sharpening method was applied to GSA(Gram-Schmidt Adaptive) and GIHS(Generalized Intensity-Hue-Saturation), which are typical CS-based algorithms, and GS2(Gram-Schmidt 2) and MTF-GLP(Modulation Transfer Functions and Generalized Laplacian Pyramid), which are typical MRA-based algorithms. In order to evaluate the performance of the sharpening methods, quantitative evaluation was performed using the quantitative indices of spectral and spatial information, and qualitative evaluation was performed through image interpretation.
As a result of evaluating the sharpening results of the Sentinel-2A satellite image, the spatial quality of the results of applying the selected band scheme based GSA algorithm was the best. In terms of spectral information, the results of sharpening of 20-m bands represented the least distortion in the results of applying synthesized band scheme based MTF-GLP. In the case of sharpening of 60-m bands, the spectral distortion was the lowest in the results of applying selected band scheme based GS2. In the sharpening results of the Worldview-3 satellite images, selected band scheme based GS2 method showed the best quality in spectral and spatial aspects. The sharpened images produced by the proposed method could be effectively applied to various fields, such as land cover classification and change detection.