Multi-aspect Polarimetric SAR Image Scattering Feature Information Coding and SVM Classification

• Published in: Asian and Pacific Conference on Synthetic Aperture Rader proceedings (APSAR) pp. 1 - 5
• Main Authors: Li, Yang, Wang, GuanYun, Zou, Yunting, Wang, YanPing, Lin, Yun, Hong, Wen
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2019
• Subjects: Anisotropic; anisotropic scattering; H/A/Alpha decomposition; Image sequences; Merging; Multi-aspect polarimetric synthetic aperture radar; Polarimetric synthetic aperture radar; Quantization (signal); Radar polarimetry; Scattering; Support vector machines; Transcoding; Vectors; Yamaguchi four-component decomposition; Yang decomposition
• ISSN: 2474-2333
• DOI: 10.1109/APSAR46974.2019.9048589

Multi-aspect polarimetric SAR images is rich of anisotropic scattering signature information for terrain classification. By using an anisotropic scattering detection model, we propose a new sequence-to-sequence signature merging framework combining aspect signatures with polarimetric decomposition signatures in this paper. First of all, the anisotropic scattering centers are detected to generate a new sequence images from the original sub-aperture image sequence. In addition, there are three polarimetric decomposition models including Yamaguchi four-component decomposition, Yang decomposition and H/A/Alpha decomposition are performed on the new sequence images to generate a series of signature curves. The framework also includes the normalization of the curve, the quantization coding and transcoding. At last, a support vector basis (SVM) classification results was used to analysis. Our algorithm was validated and applied to a set of airborne P-band fully polarimetric circular SAR data acquired by the Institute of Electronics, Chinese Academy of Science.