This study aimed to propose an improved Gram–Schmidt adaptive pansharpening method using the support vector regression (SVR) and Markov random field (MRF) models to generate the satellite image with high resolution maintaining spectral details in the cases of equal to or greater ratio than 8 between patial resolutions of low resolution multispectral (LRMS) and panchromatic (PAN) images.
The main conclusions were drawn as follows;
(1) The newly proposed GSA–SVR method using the SVR exhibited better performance than Gram–Schmidt (GS) method and its alternative methods such as Gram–Schmidt adaptive (GSA) and GSA using a polynomial regression-based injection model (GS_MLR).
(2) The quality of the initial pansharpened image obtained from the GSA–SVR was further improved by using the MRF model.
(3) The improved Gram–Schmidt adaptive pansharpening method using the SVR and MRF models produced pansharpened images efficiently in the cases of equal to or greater ratio than 8 between spatial resolutions of LRMS and PAN images.
This approach has contributed to improving land use and land cover classification because our method pansharpens LRMS image efficiently to improve its spatial resolution in the cases of high ratios between spatial resolutions of LRMS and PAN images.
Our result was published in "Journal of the Indian Society of Remote Sensing" with the title of "Improving Gram–Schmidt Adaptive Pansharpening Method Using Support Vector Regression and Markov Random Field"(https://doi.org/10.1007/s12524-024-01934-x).
