基于混合Wishart模型的极化SAR图像非监督分类

钟能; 杨文; 杨祥立; 郭威

doi:10.12000/JR16133

基于混合Wishart模型的极化SAR图像非监督分类

DOI: 10.12000/JR16133

武汉大学电子信息学院武汉 430072

基金项目:

国家自然科学基金 61331016

国家自然科学基金 61271401

详细信息

作者简介:
钟能(1993-), 男, 2015年获得吉林大学工学学士学位, 现于武汉大学电子信息学攻读硕士学位.主要研究方向为极化合成孔径雷达图像处理. E-mail: zn_whu@whu.edu.cn

杨文(1976-), 男, 教授, 博士生导师, 2004年获得武汉大学工学博士学位. 研究方向为图像处理与计算机视觉. yangwen@whu.edu.cn

杨祥立(1991-), 男, 2016年获得武汉大学工学硕士学位, 现于武汉大学电子信息学攻读博士学位.主要研究方向为极化合成孔径雷达图像处理. E-mail: xiangliyang@whu.edu.cn

郭威(1994-), 男, 2016年获得西安电子科技大学工学学士学位, 现于武汉大学电子信息学院信号处理实验室攻读硕士学位.主要研究方向为极化合成孔径雷达图像解译. E-mail: weige@whu.edu.cn

通讯作者:
杨文 yangwen@whu.edu.cn

中图分类号: TN957
计量
- 文章访问数: 2275
- HTML全文浏览量: 598
- PDF下载量: 712
- 被引次数: 0
出版历程
- 收稿日期: 2016-11-30
- 修回日期: 2017-01-16
- 网络出版日期: 2017-10-28

Unsupervised Classification for Polarimetric Synthetic Aperture Radar Images Based on Wishart Mixture Models

School of Electronic Information, Wuhan University, Wuhan 430072, China

Funds:

The National Natural Science Foundation of China 61331016

The National Natural Science Foundation of China 61271401

摘要

摘要: 极化合成孔径雷达图像非监督分类是极化SAR图像自动化解译的重要步骤，但是在非监督分类的过程中如何确定样本类数仍然是十分具有挑战性的问题。由于像素之间具有空间相关性，因此和基于像素的分类方法相比，基于区域的分类方法能得到更加鲁棒的结果。为此，该文提出了一种基于混合Wishart模型和密度峰值聚类的区域级极化SAR图像非监督分类方法。该方法首先使用SLIC算法对极化SAR图像进行过分割，生成多个超像素区域；然后采用混合Wishart模型对超像素区域进行建模，并且利用Cauchy-Schwarz散度来衡量不同超像素区域之间的距离；最后通过密度峰值快速搜索聚类算法得到PolSAR图像的非监督分类结果。在不同极化SAR图像上的实验结果表明了该文方法的有效性。
- 极化SAR图像 /
- 非监督分类 /
- 混合Wishart模型 /
- 密度峰值
Abstract: Unsupervised classification is a significant step inthe automated interpretation of Polarimetric Synthetic Aperture Radar (PolSAR) images. However, determining the number of clusters in this process is still a challenging problem. To this end, we propose a region-based unsupervised classification method for PolSAR images by introducing Wishart mixture models and a Density Peaks Clustering (DPC) algorithm. More precisely, the Simple Linear Iterative Clustering (SLIC) algorithm is first used to segment the PolSAR image into superpixels. Subsequently, the Wishart mixture models are adopted to model each superpixel, and the pairwise distances between different superpixels are measured by Cauchy-Schwarz divergence. Finally, the unsupervised classification result of the PolSAR image is obtained via clustering by fast search and find of density peaks. The experimental results obtained from different PolSAR images demonstrate that the proposed method is effective.
- Polarimetric Synthetic Aperture Radar (PolSAR) image /
- Unsupervised classification /
- Wishart mixture models /
- Density peaks

HTML全文

图 1 EMISAR数据的实验结果

Figure 1. The experiment results of EMISAR data

下载: 全尺寸图片幻灯片

图 2 AIRSAR数据

Figure 2. AIRSAR data

下载: 全尺寸图片幻灯片

图 3 AIRSAR数据的实验结果

Figure 3. The experimental results of AIRSAR data

下载: 全尺寸图片幻灯片

表 1 AIRSAR数据的分类结果

Table 1. The classification results of AIRSAR data

方法	Bartlett方法	Wishart方法	本文方法
油菜	0	0.8727	0.9647
草地	0	0.3750	0.9028
土豆	1	1	0.9879
小麦	0.9988	0.8682	0.9890
裸地	0.9934	0.9861	0.9998
豌豆	0.9584	0.9622	0.9744
甜菜	0.9696	0.0015	0.5218
大麦	0.4850	0.6471	0.9892
苜蓿	0.9571	0.9853	0.9320
OA	0.7581	0.8007	0.9420
Kappa	0.7044	0.7682	0.9336
F1-score	0.6173	0.7850	0.9337
Purity	0.7581	0.8334	0.9432

下载: 导出CSV

表 2 混合Wishart模型不同分量个数的计算时间

Table 2. The computational time of Wishart mixture models with different components

时间(s)	10	15	25	35	40
EMISAR数据	0.3885	0.4361	0.6478	0.9187	1.1588
AIRSAR数据	1.7862	2.2289	3.3989	5.0603	5.5793

下载: 导出CSV

表 3 不同步骤的计算时间

Table 3. The computational time corresponding to each step

时间(s)	SLIC	MixWishart	DPC
EMISAR数据	0.3199	0.7120	7.3037
AIRSAR数据	1.2346	3.6250	91.8320

下载: 导出CSV

参考文献(18)

[1]	Yang W, Zhong N, Yang X, et al.. Riemannian sparse coding for classification of PolSAR images[C]. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Beijing, China, 2016: 5698-5701.
[2]	田维, 徐旭, 卞小林, 等.环境一号C卫星SAR图像典型环境遥感应用初探[J].雷达学报, 2014, 3(3): 339-351. http://radars.ie.ac.cn/CN/abstract/abstract147.shtml Tian Wei, Xu Xu, Bian Xiao-lin, et al.. Application of environment remote sensing by HJ-1C SAR imagery[J]. Journal of Radars, 2014, 3(3): 339-351. http://radars.ie.ac.cn/CN/abstract/abstract147.shtml
[3]	Lee J S, Grunes M R, Ainworth T L, et al.. Unsupervised classification using polarimetric decomposition and the complex Wishart classifier[C]. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 1998, 4: 2178-2180.
[4]	Ferro-Famil L, Pottier E, and Lee J S. Unsupervised classification of multi-frequency and fully polarimetric SAR images based on the H/A/Alpha Wishart classifier[J]. IEEE Transactions on Geoscience and Remote Sensing, 2001, 39(11): 2332-2342. doi: 10.1109/36.964969
[5]	Ersahin K, Cumming I G, and Yedlin M J. Classification of Polarimetric SAR data using spectral graph partitioning[C]. IEEE International Conference on Geoscience and Remote Sensing Symposium (IGARSS), Denver, USA, 1999: 1756-1759.
[6]	Kersten P R, Lee J S, and Ainworth T L. Unsupervised classification of polarimetric synthetic aperture radar images using fuzzy clustering and EM clustering[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(3): 519-527. doi: 10.1109/TGRS.2004.842108
[7]	Song H, Yang W, Bai Y, et al.. Unsupervised classification of polarimetric SAR imagery using large-scale spectral clustering with spatial constraints[J]. International Journal of Remote Sensing, 2015, 36(11): 2816-2830. doi: 10.1080/01431161.2015.1043759
[8]	Wang Y, Han C, and Tupin F. PolSAR data segmentation by combining tensor space cluster analysis and Markovian framework[J].IEEE Geoscience and Remote Sensing Letters, 2010, 7(1): 210-214. doi: 10.1109/LGRS.2009.2031660
[9]	Rodriguez A and Laio A. Clustering by fast search and find of density peaks[J]. Science, 2014, 344: 1492-1496. doi: 10.1126/science.1242072
[10]	Tran T N, Wehrens R, Hoekman D H, et al.. Initialization of Markovian random field clustering of large remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(8): 1912-1919. doi: 10.1109/TGRS.2005.848427
[11]	Cao F, Hong W, Wu Y, et al.. An unsupervised segmentation with an adaptive number of clusters using the SPAN/H/a/A space and the complex Wishart clustering for fully Polarimetric SAR data analysis[J]. IEEE Transactions on Geoscience and Remote Sensing, 2007, 45(11): 3454-3467. doi: 10.1109/TGRS.2007.907601
[12]	Liu B, Hu H, Wang H, et al.. Superpixel-based classification with an adaptive number of classes for polarimetric SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2013, 51(2): 907-924. doi: 10.1109/TGRS.2012.2203358
[13]	Achanta R, Shaji A, Smith K, et al.. SLIC superpixel compared to state-of-the-art superpixel methods[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(11): 2274-2282. doi: 10.1109/TPAMI.2012.120
[14]	Yang W, Yang X L, Yan T H, et al.. Region-based change detection for polarimetric SAR images using wishart mixture models[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(11): 6746-6756. doi: 10.1109/TGRS.2016.2590145
[15]	Nielsen F. K-MLE: A fast algorithm for learning statistical mixture models[C]. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Kyoto, 2012: 869-872.
[16]	Nielsen F. Closed-form information-theoretic divergences for statistical mixtures[C]. International Conference on Pattern Recognition, Tsukuba, 2012: 1723-1726.
[17]	谢娟英, 高红超, 谢维信. K近邻优化的密度峰值快速搜索聚类算法[J].中国科学:信息科学, 2016, 46(2): 258-280. http://kns.cnki.net/KCMS/detail/detail.aspx?filename=pzkx201602008&dbname=CJFD&dbcode=CJFQ Xie J Y, Gao H C, and Xie W X. K-nearnestneighbors optimized clustering algorithm by fastsearch and finding the density peaks of a dataset[J]. Scientia Sinica Informationis, 2016, 46(2): 258-280. http://kns.cnki.net/KCMS/detail/detail.aspx?filename=pzkx201602008&dbname=CJFD&dbcode=CJFQ
[18]	Cherian A, Morellas V, and Papanikolopoulos N. Bayesian nonparametric clustering for positive definite matrices[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(5): 862-874. doi: 10.1109/TPAMI.2015.2456903