Jinghong Han, Haijiang Wang, Mengqing Gao, and Min Sun
Polarimetric SAR, LLE algorithm, majority voting, Wishart, super-pixel, classification
To improve the classification accuracy of polarimetric synthetic aperture radar (PolSAR) images, a classification of algorithm based on superpixel is proposed, and the locally linear embedding (LLE) dimension reduction algorithm is improved in the process of reducing the feature dimension. The traditional image classification methods are based on pixel, and the classification effect is not satisfactory. The classification method proposed in this paper is based on superpixel segmentation combined with majority voting algorithm and Wishart algorithm. This method is superior to traditional algorithms. The LLE algorithm is improved, and the distance metric combining the Wishart distance with the Euclidean distance is proposed. This method makes the dimension reduce data more favourable for classification. Experimental results of two PolSAR images are presented in this paper. The results show that the proposed method is superior to the traditional method and can achieve better classification effect.
Important Links:
Go Back
