Fusion of polarimetric and texture information for urban building extraction from fully polarimetric SAR imagery

Building extraction from remote sensing images is very important in many fields, such as urban planning, land use investigation, damage assessment, and so on. In polarimetric synthetic aperture radar (PolSAR) imagery, the buildings not only have typical polarimetric features but also have rich texture features. In this paper, the texture information is introduced to improve the accuracy of urban building extraction from PolSAR imagery by a new method called cross reclassification. Based on this method, the polarimetric information-based results and texture-based results can be effectively fused. The experimental results of three representative PolSAR images with different characteristics demonstrate the effectiveness of the proposed method, and the accuracy of building extraction can be improved, compared with the traditional method using only polarimetric information.