A knowledge-based procedure for remote sensing image classification

Abstract

Object-based image analysis (OBIA) in combination with domain knowledge is an important methodology for remote sensing image interpretation. It can deal with semantic gap effectively. In this paper, a knowledge-based procedure for remote sensing image classification is presented. Domain knowledge is represented with semantic network, which guides the initial classification of the image. A quantitative matching degree is used in the modification of the initial classification. The proposed procedure is much more efficient compared with method in literature.