Investigation of extreme reflections of metal ceilings and salty soils using object oriented satellite image processing Sentinel-2 L1C using SVM classification method

The Sentinel-2 provides available multispectral bands at relatively high spatial resolution but does not acquire the panchromatic band. In this study, using Sentinel-2 images, the reflectance of metal roofs has been investigated and the differences between these reflections with other high reflections such as saline and dry soils have been evaluated. Bands 2, 3, 4 and band 8, which have a resolution of ten meters, are the most used in extracting different types of reflection. The result of the research shows that using the reflection of materials, it is easy to identify and harvest samples for the purpose of classifying the controlled sample by object-oriented processing. The results show that there is a significant difference between the reflection of the salty soil and the metal roof in the near infrared range, although in the image with the natural color combination, both types of material show a type and a reflection. This paper presents a new approach for extracting training samples from metal roofs compared to saline soils. The classification of SVM as the best method of classification with an accuracy of 96.9% and Kappa coefficient of 0.9 for categorization in this study was selected among other classification methods.