Classification of satellite images is possible by the use of two scientific methods; the method of “basic pixel” which is based on classification of numerical values of images and the method “object oriented” which in addition to numerical values, uses data related to the content, texture and background in the classification process of images. The purpose of this study is to compare the classification algorithm of the maximum likelihood in the based pixel processing method with the closest neighbor algorithm in the object oriented processing for evaluation of classification performance of Landsat 8 satellite images. In order to compare the results, the land use map of Malayer city in Iran as an important place of the country for agricultural production and grape industry, was prepared by both methods of classification. The results indicate that with a 16% increase in precision of both overall accuracy and kappa for classification of satellite images, the object oriented method has a higher accuracy.
