Comparative study of image fusion methods for WorldView-3 data

WorldView-3 is one of the highest spatial resolution multispectral commercial satellites currently available on the market, with great potential for applications in land analysis, geological mineral surveys, and disaster assessments. In order to fully utilize the high spatial resolution of WorldView-3 data, fusion processing was performed using Principal Component Analysis (PCA), Gram-Schmidt (G-S), Histogram Contrast Stretching (HCS), and Nearest-Neighbor Diffusion (NNDiffuse) methods to combine the multispectral images with the panchromatic image. Qualitative and quantitative evaluations were conducted on the fused images, and the results showed that in the visible-near infrared (VNIR) wavelength range, the HCS method had the best overall fusion effect, with the closest mean value to the original multispectral image, highest standard deviation, information entropy, and average gradient value. The NNDiffuse method had the highest correlation coefficient and spectral fidelity between the fused image and the original multispectral image. In the shortwave infrared wavelength (SWIR) bands, the HCS fused image had the best spectral and color fidelity, while the NNDiffuse and G-S methods produced fused images with richer information and hierarchy, higher clarity, and more distinct spatial structure and texture features.