Comparing the performance of parametric and nonparametric methods of land cover classification by using of Landsat satellite images (Case study:a part of Dezful township)

Nowadays، remote sensing data is able to provide the latest information for the study of land cover and land uses. These images are important to provide land use maps، because of present update information، diversity of forms، digital and also ability of images processing. Specifying of Land cover helps to managers for its make decision in different situation. Different methods are applicable in classifying and presenting land use maps and land cover stem from satellite image. These methods have variable functions and outputs depending on type and characteristics of uses data. The purpose of this study is comparing of Parametric (including distance functions and box functions) and nonparametric (SVM) methods in classification of land cover and land use using multiband images of Landsat satellite 8. This study is applied-developmental، and is done in descriptive-analytical method. The data includes satellite images of Landsat satellite 8 (13. 08. 2013) that was prepared، then using ENVI software were prepared and analyzed. The effectiveness of each classification method were investigated whit calculate of Overall accuracy and Kappa coefficient indicates. The results show that the SVM algorithm، especially linear، radial and polynomial Kernel than parametric methods with 97. 15%، 95. 89% and 95. 63%، respectively، have more accuracy. This study confirms the performance of SVM algorithm in comparison whit parametric methods.