Artificial cover extraction based on a Hierarchical Stripping Model in the Loess Plateau, China

Abstract

This paper proposes a Hierarchical Stripping Model (HSM) to extract artificial cover in the Loess Plateau of China by stripping other no-artificial covers (e.g. water, vegetable, cropland, bare) hierarchically. Firstly, a Statistic Divisibility Analysis (SDA) is established to evaluate the divisibility between artificial and no-artificial cover and the divisibility values are the key base of specifying an optimal stripping sequence. And then, each no-artificial class with distinct level of divisibility is stripped by different ways which includes artificial cover index, Support Vector Machines (SVM) classification, object-oriented expert knowledge and object-oriented post-classification. This method was developed and tested on one Landsat path/raw study site that contain Yan'an City, and the overall accuracy and Kappa coefficient of the study area were 98.9286% and 0.9786 respectively. Therefore, the method has the potential to provide a robust method to extract artificial cover in complex large area.