A method of salt-affected soil information extraction based on a support vector machine with texture features

This paper describes an effort to apply an improved support vector machine classifier to classify salt-affected soil. In this study, we used the support vector machine with texture features to extract thematic information for salt-affected soil. The SVM classification was conducted using a combination of multi-spectral features and texture features as the data source. We used mean, variance and homogeneity features, which were the best texture features, to improve the classification. In addition, we provided a contrast between the proposed SVM method and other SVM methods. The results revealed that the SVM classification used here can effectively extract salinization soil thematic information for the Yinchuan Plain. Specifically, the accuracy of this method was 84.6974% and the kappa coefficient was 0.8202, which indicated superiority over other classification methods.