Preparing landuse and vegetation cover in planning and management of natural resources is very important. For this aim, using remote sensing data has important role especially because of daily cover and low cost images. So in this research, Landsat 8 images are used as input data for extracting landuse ...
Read More
Preparing landuse and vegetation cover in planning and management of natural resources is very important. For this aim, using remote sensing data has important role especially because of daily cover and low cost images. So in this research, Landsat 8 images are used as input data for extracting landuse map in levels 1 and 2. Images with respect to newly issuing are correcting radiometric by using relationships in ERDAS software's modeling formulation environment. Also, NDVI , BI and PCA as input beside other bands are used for increasing classification. SVM is evaluated and it’s the best result compared with ANN. Results have been shown SVM with 92% accuracy with Kappa Index 0.91 and ANN with 89% accuracy with Kappa Index 0.87. SVM have better result than ANN in all places which classes have same behavior.
