The application of rough set theory in remote sensing image classification

Rough set theory is a relatively new mathematical tool to deal with imprecise, incomplete and inconsistent data. The content of this paper is twofold. First, to decrease computational time, band reduct is performed on multispectral remote sensing image using rough set and information entropy. Second, image classification is obtained after band reduction, Gaussian mixture model and EM algorithm are considered. The algorithm designed in this paper can make bands reduction and multispectral remote sensing image classification unsupervised. Experimental results show that the proposed method did have effective and valid performance.