Comparing pixel-based and object-based algorithms for classifying land use of arid basins (Case study: Mokhtaran Basin, Iran)

Abstract

In this research, two techniques of pixel-based and object-based image analysis were investigated and compared for providing land use map in arid basin of Mokhtaran, Birjand. Using Landsat satellite imagery in 2015, the classification of land use was performed with three object-based algorithms of supervised fuzzy-maximum likelihood, maximum likelihood, and K-nearest neighbor. Nine combinations were examined in terms of scale level (SL10, SL30, and SL50) and the nearest neighborhood (NN3, NN5, and NN7) in an object-based classification. Ultimately, the validity was evaluated through the usage of two disagreement components including allocation disagreement and quantity disagreement. Results of maximum likelihood classification showed higher overall inaccuracy compared to images categorized based on fuzzy-maximum likelihood and object-based nearest neighbor algorithms. The SL30-NN3 object-based classifier decreased the quantity disagreement by 290% compared to the maximum likelihood and 265% compared to fuzzy-maximum likelihood classifiers. For allocation disagreement, these values were equal to 36% and 19%, respectively. Thus, object-based classification had a better performance in land-use classification of Mokhtaran basin.