Evaluation of optical and synthetic aperture radar image fusion methods: a case study applied to Sentinel imagery

Abstract

This paper evaluates different optical and synthetic aperture radar (SAR) image fusion methods applied to open-access Sentinel images with global coverage. The objective of this research was to evaluate the potential of image fusion methods to get a greater visual difference in land cover, especially in oil palm crops with natural forest areas that are difficult to differentiate visually. The application of the image fusion methods: Brovey (BR), high-frequency modulation (HFM), Gram-Schmidt (GS), and principal components (PC) was evaluated on Sentinel-2 optical and Sentinel-1 SAR images using a cloud computing environment. The results show that the application of the implemented optical/SAR image fusion methods allows the creation of a synthetic image with the characteristics of both data sources. The multispectral information provided by the optical image and information associated with the geometry and texture/roughness of the land covers, provided by the SAR image, allows a greater differentiation in the visualization of the various land covers, achieving a better understanding of the study area. The fusion methods that visually presented greater characteristics associated with the SAR image were the BR and GS methods. The HFM method reached the best statistical indicators; however, this method did not present significant visual changes in the SAR contribution.