The study of the Vegetation is a crucial factor in the ecosystem. This study investigates the improved classification for urban vegetation on Giglio Island (in Italia) by integrating data from Sentinel-1 (Synthetic Aperture Radar) and Sentinel-2 (Multispectral) from the European Space Agency's Sentinel constellation. The analysis leverages the power of Random Forest Algorithm and Sklearn python’s library to create a classification map. Usually, urban vegetation mapping relies on a single data source resulting in limitations in accuracy and the abilities to differentiate vegetation types. The vegetation treated were: Mediterranean macchia, grasslands, coastal vegetation, pine forest. First, we performed independent classification vegetation using Normalized Difference Vegetation Index and Radar Vegetation Index. Secondly, to enhance classification accuracy, by incorporating a combination of each index with the Modified Normalized Difference Water Index and Soil Adjusted Vegetation Index we identified the vegetation classes exhibiting the highest prevalence. Thirdly using the proposed fusion method we combined the Modified Normalized Difference Water Index and Soil Adjusted Vegetation Index with the fusion between Normalized Difference Vegetation Index and Radar Vegetation Index and we compared the density variations among vegetation types. The results showed the effectiveness of the fusion approach significantly improved the classification accuracy, a perfect 100% of overall accuracy and Kappa index in all prediction elements. It showed a classification result by class for Random Forest Algorithm where each vegetation types were “present”, and showed the agreement level by class for Random Forest Algorithm some vegetation types were perfect, moderate, fair, slight, poor, very poor, inexistent.