15 October 2015 A lithological classification method from fully polarimetric SAR data using Cloude-Pottier decomposition and SVM
Author Affiliations +
Proceedings Volume 9674, AOPC 2015: Optical and Optoelectronic Sensing and Imaging Technology; 967405 (2015) https://doi.org/10.1117/12.2196856
Event: Applied Optics and Photonics China (AOPC2015), 2015, Beijing, China
Abstract
This article puts forward a kind of lithological classification method to take advantage of the fully polarimetric SAR data for lithological classification by the combination of cloude-pottier decomposition and support vector machine(SVM). Cloude-pottier target decomposition method is used to extract three characteristic parameters from the fully polarimetric SAR data as polarization entropy(H), scattering Angle(α), and the anisotropic(A) in xingcheng region, Liaoning province. And these parameters are taken as a sample vector and selected as the radial basis function for the SVM classifier. Thus the lithological classification from the fully polarimetric SAR images is implemented for the study area. By the comparation to the geological map, the classification results can consist with the actual rock distribution very well, and the overall classification precision reaches 80.0871%. But wishart supervised classification precision reaches 73.3837% , It shows that the method is feasible and effective for full polarization SAR image classification. Compared with the conventional classification method, it greatly improves the accuracy of interpretation.
© (2015) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Minghui Xie, Minghui Xie, Qi Zhang, Qi Zhang, Shengbo Chen, Shengbo Chen, Fengli Zha, Fengli Zha, } "A lithological classification method from fully polarimetric SAR data using Cloude-Pottier decomposition and SVM", Proc. SPIE 9674, AOPC 2015: Optical and Optoelectronic Sensing and Imaging Technology, 967405 (15 October 2015); doi: 10.1117/12.2196856; https://doi.org/10.1117/12.2196856
PROCEEDINGS
8 PAGES


SHARE
Back to Top