Urban classification utilizing SVM and machine learning techniques on Sentinel-۲ imagery in Tabriz, Iran

The monitoring of urban land using satellite imagery has advanced significantly due to modern technological innovations and the growing number of satellites in orbit. Current studies primarily utilize Sentinel-۲ satellite images from the European Space Agency (ESA). For supervised classification, both maximum likelihood (ML) and support vector machine (SVM) algorithms have been applied. To enhance the accuracy of the supervised classification for Sentinel-۲ data from November ۱۴, ۲۰۲۳, texture analysis using the gray-level co-occurrence matrix (GLCM) was implemented for slant images. The classification outcomes indicated that the map generated by the SVM algorithm, which achieved a kappa coefficient of ۰.۹۰, outperformed the ML algorithm, which had a kappa coefficient of ۰.۸۶. In terms of producer and user accuracy, the SVM demonstrated superior performance across five classes. The ML algorithm performed well in distinguishing water bodies from other categories. In regions with wider streets, classification was straightforward; however, in narrower streets, backscattering to some pixels being incorrectly classified into other categories.

کلیدواژه ها:

maximum likelihood ، support vector machine ، GLCM ، classification ، Tabriz