采用联合相似度谱聚类集成的极化SAR分类  

Combined similarity based spectral clustering ensemble for POLSAR classification

作　　者：刘璐[1] 王蓉芳[1] 焦李成[1] 石俊飞[1] 

机构地区：[1]西安电子科技大学智能感知与图像理解教育部重点实验室,陕西西安710071

出　　处：《西安电子科技大学学报》2015年第3期48-53,128,共7页Journal of Xidian University

基　　金：国家重点基础研究发展计划资助项目(2013CB329402);国家自然科学基金资助项目(61072106;61272282)

摘　　要：为了降低谱聚类算法尺度参数对分类结果的影响,避免Nystrm逼近导致的分类结果不稳定,提出了一种基于谱聚类集成的极化合成孔径雷达(SAR)地物分类方法.首先,利用像素间的空间关系和极化信息,将基于相干矩阵Wishart距离的相似性测度和基于极化特征矢量欧氏距离的相似性测度相结合,引入马尔可夫随机场势函数,构造谱聚类的相似性矩阵;然后,采用基于Nystrm逼近的谱聚类实现极化SAR数据的单次谱分类;最后,采用集成策略完成对极化SAR图像的地物分类.实验结果表明,该算法提高了分类精度,区域一致性保持较好,且分类结果稳定.In order to improve the robustness of spectral clustering to the scaling parameter and avoid the instable results caused by the Nystr6m approximation, a novel spectral clustering ensemble method for Polarimetric SAR （PolSAR） land cover classification is proposed. Firstly, Wishart-derived distance measure and polarimetric similarity are combined to obtain the complementary information from the spatial and polarimetric relations between pairwise pixels. The Markov Random Field （MRF） potential function is introduced to construct the similarity matrix. Then the Nystr6m approximation based spectral clustering is employed to achieve a single spectral classification of PolSAR data. Finally, multiple individual classifications are obtained and integrated by an ensemble strategy. Experimental results demonstrate that the proposed method improves the classification performance and region harmony, and leads to stable results.

关 键 词：极化合成孔径雷达 谱聚类集成 Wishart距离 马尔可夫随机场 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]