Buildings Extraction from Polarimetric SAR Image Using Improved Three-component Decomposition and Wishart Classification

• Main Authors: LIU Xiuguo, JIANG Ping, CHEN Qihao, CHEN Q
• Format: Article
• Language: zho
• Published: Surveying and Mapping Press 2015-02-01
• Series: Acta Geodaetica et Cartographica Sinica
• Subjects: polarimetric SAR; buildings extraction; three-component decomposition; selective de-orientation
• Online Access: http://xb.sinomaps.com:8081/Jwk_chxb/CN/10.11947/j.AGCS.2015.20130535

To address the misclassification issue on buildings extraction based on Freeman decomposition method, a novel improved three-component decomposition model is proposed in this paper. By combining the selective de-orientation derived from the circular polarization correlation coefficient method with the generalized volume scattering model, it can accurately characterize the scattering characteristics of surface features. On this basis, the complex Wishart iterative classification is introduced to develop a new method of buildings extraction. An E-SAR L band polarimetric SAR image was used to verify the effectiveness of this modified algorithm. The experiment result shows it could perform better in distinguishing between oblique buildings and forest, and consequently improve the accuracy of buildings extraction.