Extraction of Buildings with Multi-features from Polarization Synthetic Aperture Radar Image
It is of great significance to extract buildings from Polari-metric Synthetic Aperture Radar (PolSAR) images for disas-ter reduction and prevention. This paper makes full use of the complementarities between different polarization features and texture features to extract buildings. First,extracting the po-larization features and texture features. Then,random forest method with features selection is used to extract buildings and analyze the importance of features. In this study,airborne AIRSAR and ESAR data are used for experiments. The re-sults show that the overall accuracy of the proposed method is higher,reaching 97.77％ and 97.01％,respectively. In addi-tion,in PolSAR images,the polarization feature is the main feature,while the texture feature is the secondary feature,and the texture feature can be used as a supplement to the polariza-tion feature.
building extractionPolSARpolarimetric featurestexture featuresRF
万方数据
维普
