基于聚類識(shí)別的極化SAR圖像分類
doi: 10.11999/JEIT180229 cstr: 32379.14.JEIT180229
-
1.
西安電子工程研究所 ??西安 ??710100
-
2.
西安交通大學(xué)電子與信息工程學(xué)院 ??西安 ??710049
PolSAR Image Classification Based on Discriminative Clustering
-
1.
Xi’an Electronic Engineering Research Institute, Xi’an 710100, China
-
2.
School of Electronic and Information Engineering, Xi’an Jiaotong University, Xi’an 710049, China
-
摘要: 該文提出一種基于判別式聚類框架的非監(jiān)督極化SAR圖像分類算法，利用判別式監(jiān)督分類技術(shù)實(shí)現(xiàn)非監(jiān)督聚類。為實(shí)現(xiàn)該算法，定義了一個(gè)結(jié)合softmax回歸模型和馬爾科夫隨機(jī)場光滑性約束的能量函數(shù)。該模型中，像素類標(biāo)和分類器均為需要優(yōu)化的未知變量。該算法從基于
${H / {\bar \alpha }}$ 目標(biāo)極化分解和K-Wishart極化統(tǒng)計(jì)分布而產(chǎn)生的初始化類標(biāo)開始，交替迭代優(yōu)化分類器和類標(biāo)的能量函數(shù)，從而實(shí)現(xiàn)對(duì)分類器和類標(biāo)的求解。真實(shí)極化SAR數(shù)據(jù)上的實(shí)驗(yàn)結(jié)果證明了該算法的有效性和先進(jìn)性。-
關(guān)鍵詞:
- 極化SAR圖像分類 /
- 判別式聚類 /
- 馬爾科夫隨機(jī)場 /
- softmax回歸模型
Abstract: This paper presents a novel unsupervised image classification method for Polarimetric Synthetic Aperture Radar (PolSAR) data. The proposed method is based on a discriminative clustering framework that explicitly relies on a discriminative supervised classification technique to perform unsupervised clustering. To implement this idea, an energy function is designed for unsupervised PolSAR image classification by combining a supervised Softmax Regression (SR) model with a Markov Random Field (MRF) smoothness constraint. In this model, both the pixelwise class labels and classifiers are taken as unknown variables to be optimized. Starting from the initialized class labels generated by Cloude-Pottier decomposition and K-Wishart distribution hypothesis, the classifiers and class labels are iteratively optimized by alternately minimizing the energy function with respect to them. Finally, the optimized class labels are taken as the classification result, and the classifiers for different classes are also derived as a side effect. This approach is applied to real PolSAR benchmark data. Extensive experiments justify that the proposed approach can effectively classify the PolSAR image in an unsupervised way and produce higher accuracies than the compared state-of-the-art methods. -
表 1 本文算法所使用的特征
極化特征分類 標(biāo)識(shí) 物理描述 極化矩陣及其數(shù)學(xué)變換 ${{{T}}_{ij}}(i,j = 1,2,3,i \le j)$ 水平垂直線極化方式下的相干矩陣元素(模值與幅角) ${\rm{Lin}}45{{{T}}_{ij}}(i,j = 1,2,3,i \le j)$ +45°/–45°線極化方式下的相干矩陣元素(模值與幅角) ${\rm{Cir}}45{{{T}}_{ij}}(i,j = 1,2,3,i \le j)$ 左右旋圓極化方式下的相干矩陣元素(模值與幅角) $\frac{{{I_{{\rm{hv}}}}}}{{{I_{{\rm{hh}}}}}},\frac{{{I_{{\rm{hv}}}}}}{{{I_{{\rm{vv}}}}}},\frac{{{I_{{\rm{hh}}}}}}{{{I_{{\rm{vv}}}}}},\frac{{{I_{{\rm{rr}}}}}}{{{I_{{\rm{lr}}}}}},\frac{{{I_{{\rm{ll}}}}}}{{{I_{{\rm{lr}}}}}},\frac{{{I_{{\rm{ll}}}}}}{{{I_{{\rm{rr}}}}}},\frac{{{I_{{\rm{mn}}}}}}{{{I_{{\rm{mm}}}}}},\frac{{{I_{{\rm{mn}}}}}}{{{I_{{\rm{nn}}}}}},\frac{{{I_{{\rm{mm}}}}}}{{{I_{{\rm{nn}}}}}}$ 水平垂直線極化，+45°/–45°線極化以及
左右旋圓極化方式下的強(qiáng)度比值SPAN 極化總功率 目標(biāo)分解特征 Pauli矩陣分解 Pauli分解參數(shù) ${P_{\rm{s}}},{P_{\rmq7j3ldu95}},{P_{\rm{v}}},{\alpha _{\rm{L}}}$ Freeman分解參數(shù) $\bar \alpha ,H,A,\beta ,(1 - H)(1 - A),(1 - H)A,H(1 - A),HA$ ${H / {\bar \alpha }}$分解參數(shù) 
下載: 導(dǎo)出CSV
表 2 不同方法在Flevoland地區(qū)數(shù)據(jù)上的分類準(zhǔn)確率(%)
方法/類別 空地 大麥 苜蓿 豌豆 土豆 甜菜 小麥 總分類準(zhǔn)確率 $H/\bar \alpha {\scriptsize{-}} {\rm{Wishart}}$ 99.94 86.60 84.63 88.50 81.44 84.00 82.48 85.59 Wishart MRF 100.00 86.21 93.27 94.24 85.55 89.54 92.55 91.82 Freeman Wishart 98.26 87.97 83.38 93.22 98.13 92.20 84.53 90.16 功率熵和共極化率 99.44 89.85 75.22 85.92 85.79 91.01 87.89 88.30 Wishart TMF 99.81 97.60 86.10 98.27 90.00 95.86 97.33 95.86 本文算法 100.00 100.00 95.07 98.21 98.57 98.63 99.93 99.05
下載: 導(dǎo)出CSV
表 3 本文方法在Oberpfaffenhofen地區(qū)數(shù)據(jù)上的分類準(zhǔn)確率(%)
對(duì)象/類別 森林 道路 郊區(qū) 開闊地 總分類準(zhǔn)確率 ${H / {\bar \alpha }} {\scriptsize{-}} {\rm{Wishart}}$ 65.21 41.80 51.51 76.01 64.43 Wishart MRF 74.43 55.52 42.28 72.93 67.26 Freeman Wishart 97.71 18.13 16.80 87.85 73.35 本文算法 93.65 64.70 60.78 84.41 82.25
下載: 導(dǎo)出CSV
-
KONG J A, SWARTZ A A, YUEH H A, et al. Identification of terrain cover using the optimum polarimetric classifier[J]. Journal of Electromagnetic Waves and Applications, 1988, 2(2): 171–194. POTTIER E and SAILLARD J. On radar polarization target decomposition theorems with application to target classification by using network method[C]. Proceedings of the International Conference on Antennas and Propagation, York, UK, 1991, 1: 265–268. FUKUDA S and HIROSAWA H. Support vector machine classification of land cover: Application to polarimetric SAR data[C]. Proceedings of the IEEE International Conference on Geoscience and Remote Sensing Symposium, Sydney, NSW, Australia, 2001, 1: 187–189. JIAO Licheng and LIU Fang. Wishart deep stacking network for fast PolSAR image classification[J]. IEEE Transactions on Image Processing, 2016, 25(7): 3273–3286 doi: 10.1109/TIP.2016.2567069 ZHOU Yu, WANG Haipeng, XU Feng, et al. Polarimetric SAR image classification using deep convolutional neural networks[J]. IEEE Geoscience Remote Sensing Letters, 2017, 13(12): 1935–1939 doi: 10.1109/LGRS.2016.2618840 GAO Wei, YANG Jian, and MA Wenting. Land cover classification for polarimetric SAR images based on mixture models[J]. Remote Sensing, 2014, 6(5): 3770–3790 doi: 10.3390/rs6053770 CLOUDE S R and POTTIER E. An entropy based classification scheme for land application of polarimetric SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 1997, 35(1): 68–78 doi: 10.1109/36.551935 FREEMAN A and DURDEN S L. A three-component scattering model for polarimetric SAR data[J]. IEEE Transactions on Geoscience and Remote Sensing, 1998, 36(3): 963–973 doi: 10.1109/36.673687 YAMAGUCHI Y, MORIYAMA T, ISHIDO M, et al. Four-component scattering model for polarimetric SAR image decomposition[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(8): 1699–1706 doi: 10.1109/TGRS.2005.852084 XU Feng and JIN Yaqiu. Deorientation theory of polarimetric scattering targets and application to terrain surface classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(10): 2351–2364 doi: 10.1109/TGRS.2005.855064 DOULEGERIS A P, ANFINSEN S N, and ELROFT T. Classification with a non-Gaussian model for PolSAR data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2008, 46(10): 2999–3009 doi: 10.1109/TGRS.2008.923025 徐豐, 王海鵬, 金亞秋. 深度學(xué)習(xí)在SAR目標(biāo)識(shí)別與地物分類中的應(yīng)用[J]. 雷達(dá)學(xué)報(bào), 2017, 6(2): 136–148 doi: 10.12000/JR16130XU Feng, WANG Haipeng and JIN Yaqiu. Deep Learning as Applied in SAR Target Recognition and Terrain Classification[J]. Journal of Radars, 2017, 6(2): 136–148 doi: 10.12000/JR16130 鐘能, 楊文, 楊祥立, 等. 基于混合Wishart模型的極化SAR圖像非監(jiān)督分類[J]. 雷達(dá)學(xué)報(bào), 2017, 6(2): 136–148 doi: 10.12000/JR16133ZHONG Neng, YANG Wen, YANG Xiangli, et al. Unsupervised Classification for Polarimetric Synthetic Aperture RadarImages Based on Wishart Mixture Models[J]. Journal of Radars, 2017, 6(2): 136–148 doi: 10.12000/JR16133 BACH F and HARCHAOUI Z. DIFFRAC: A discriminative and flexible framework for clustering[C]. Proceedings of Conference and Workshop on Neural Information Processing Systems, Vancouver, British Columbia, Canada, 2007: 49–56. SUN Jian and PONECE J. Learning discriminative part detectors for image classification and cosegmentation[C]. Proceedings of IEEE International Conference on Computer Vision, Sydney, NSW, Australia, 2013: 3400–3407. ZHU Ciyou, BYRD R H, LU Peihuang, et al. L-BFGS-B: Fortran subroutines for large-scale bound-constrained optimization[J]. ACM Transaction Mathematical Software, 1997, 23(4): 550–560 doi: 10.1145/279232.279236 WU Yonghui, JI Kefeng, YU Wenxian, et al. Region-based classification of polarimetric SAR images using Wishart MRF[J]. IEEE Geoscience and Remote Sensing Letters, 2008, 5(4): 668–672 doi: 10.1109/LGRS.2008.2002263 LEE J S, GRUNES M R, Ainsworth T L, et al. Unsupervised classification using polarimetric decomposition and the complex Wishart classifier[J]. IEEE Transactions on Geoscience and Remote Sensing, 1999, 37(5): 2249–2258 doi: 10.1109/36.789621 WANG Shuang, LIU Kun, PEI Jingjing, et al. Unsupervised classification of fully polarimetric SAR images based on scattering power entropy and copolarized ratio[J]. IEEE Geoscience and Remote Sensing Letters, 2013, 10(3): 622–626 doi: 10.1109/LGRS.2012.2216249 LIU Gaofeng, LI Ming, WU Yan, et al. PolSAR image classification based on Wishart TMF with specific auxiliary field[J]. IEEE Geoscience and Remote Sensing Letters, 2014, 11(7): 1230–1234 doi: 10.1109/LGRS.2013.2290066 LEE J S, GRUMES M R, POTTIER E, et al. Unsupervised terrain classification preserving polarimetric scattering characteristics[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(4): 722–731 doi: 10.1109/TGRS.2003.819883 -
圖(6) / 表(3)
計(jì)量
- 文章訪問數(shù): 3129
- HTML全文瀏覽量: 925
- PDF下載量: 143
- 被引次數(shù): 0
下載:
