Geographic unit sequences based remote sensing image analysis: A case study of Sichuan, China

摘要

Based on the analysis of the existing problems in pixel-based image processing and object based image analysis (OBIA), this paper proposed a Geographic Unit Sequences (GUS) based image analysis approach, designed to apply geographic ontology knowledge to object based image analysis, so that the image analysis results become the carriers of geographic principle rather than the distribution of land cover. This method consists of two steps: one is the construction of GUS, introducing the geographic ontology knowledge and specialist experience knowledge to guide the image analysis process; the other is the hybrid object based image analysis of bottom-up data-driven way and up-bottom knowledge-driven way. The proposed analysis method proved to be of high accuracy by the application in Xiang'E, Sichuan area with high resolution RS images. Finally, we summarized the advantages of the method verified by the application and give certain recommendations concerning practical use of proposed method.