Özet:

Abstract In order to effectively describe land cover, scientists, academics, and researchers developed machine learning classification algorithms. According to studies, these classification techniques perform better than more tried-and-true conventional techniques. The primary aim of this project is to determine the most effective strategy for categorizing land cover in order to retrieve data from Wasit. The Maximum Likelihood Classifier (MLC), which is based on the neighborhood function, and the Support Vector Machine (SVM), which is based on the ideal hyper-plane function, are two supervised classification techniques that are contrasted using Sentinel-2 data. Four land cover classes have been chosen for this optimization. Four spatial layers of the research region were surveyed with the aim of collecting and providing field-based training samples. The error matrix and kappa statistics have been used to evaluate the accuracy of each classifier. Results demonstrated that SVM performs superior to MLC. SVM and MLC have overall accuracies of 99.79 and 99.60%, respectively, and kappa coefficients of 0.997 and 0.994.

Anahtar Kelimeler: