An evaluation of pansharpening algorithms on Worldview-4 satellite imagery over Western Himalaya

This study compares component substitution (CS) and multiresolution analysis (MRA) pansharpening algorithms applied to high-resolution WorldView-4 imagery over the Indian Western Himalaya. The performance of these methods was evaluated using quantitative (i.e., visual assessment) and qualitative metrics (such as Relative Average Spectral Error (RASE), Root Mean Square Error (RMSE), Error Relative Global Dimensionless Synthesis (ERGAS), Bias, and the Fidelity-Deformation (FD) metric). The FD metric captures both spectral fidelity and spatial structure preservation by integrating localized and global error measures. The results indicated that MRA-based approaches (i.e., ATWT_M2, M3, and MTF_GLP) exhibit reduced spectral distortions, as reflected by lower Bias and RASE values, making them suitable for applications that demand high spectral fidelity. In contrast, CS-based approaches, such as HCS and BDSD, achieved lower ERGAS and RMSE values, suggesting improved spatial detail preservation. Overall, although pansharpened imagery may be advantageous for developing fine-resolution applications, the choice of the pansharpening algorithm should be made carefully, considering the specific application.