Interaction of image fusion techniques and atmospheric correction for improve SVM accuracy

Changes in land cover (LC) influence the geological, climatic, and biological environments of the earth. A precise LC map provides detailed information on resource management and intergovernmental collaboration to address global warming and biodiversity reduction. This article investigated the influence of atmospheric correction and four fusion techniques (Brovey, Hue-Saturation-Value (HSV), Principle Components (PC), and Gram-Schmidt (GS) spectral sharpening) on LC classification accuracy. Support Vector Machine (SVM) and Maximum Likelihood Classifier (MLC) were used as LC mapping classification algorithms using panchromatic (PAN) and multispectral (MS) Landsat 8 OLI satellite imagery. This study indicated that the SVM technique led to greater accuracy than MLC by 17%, 4.48%, 4.4%, 3.97%, and 15.57% in Overall Accuracy (OA) for HSV, Brovey, PC, GS, and MS images alternatively. The GS sharpened image gave the highest outcome across all SVM categories by 85.14% with a Kappa coefficient (K) value of 0.81 compared with the GS-MLC classification of about 4% in OA and 0.04 in K value. The OA distinction between HSV, Brovey, PC, GS, and MS images with and without atmospheric correction was also about 5% on SVM classification occasions which is consistent with MLC outcomes. These findings have shown that atmospheric correction is not required for LC mapping based on DN image Nevertheless, fusion performs a more influential role in the accuracy of classification than atmospheric correction. A combination of atmospheric correction, pansharpening, and SVM could offer promising LC maps from Landsat 8 OLI multispectral and panchromatic images.