摘要
合成孔径雷达(Synthetic Aperture Radar,SAR)作为一种信息获取手段,在国防、环境等方面具有突出的战略意义。SAR图像地物分类是SAR图像解译的重要内容,基于马尔可夫随机场(Markov Random Field,MRF)和统计建模的SAR图像地物分类方法充分利用了图像中的上下文信息和图像灰度统计分布特征,在图像处理中得到了广泛的应用。本文系统研究了基于MRF和统计建模的SAR图像地物分类方法。
首先,作为全文研究的理论基础,深入分析了SAR图像杂波的统计模型。文中全面概括了现有SAR图像杂波统计模型,深入研究了基于Mellin变换的参数估计方法,总结了该参数估计方法在已有分布模型中的参数求解方程。为了满足高分辨率条件下的数据建模需求,文中重点研究分析了一种基于字典集的SAR图像统计建模方法(DSEM)。为了完成多极化SAR数据建模,研究了多通道数据建模工具:copula连接函数。
其次,研究了基于MRF模型和统计建模的单极化SAR图像分类方法。文中研究了MRF模型应用于图像分类的理论框架,并结合SAR图像统计建模基础知识,研究了SAR图像分类方法以及求解最优标记矩阵的优化算法。基于copulas理论提出一种首先建立灰度数据空间和对比度纹理数据空间的联合分布,然后基于MRF模型完成单极化图像分类的方法。
最后,研究了基于MRF模型和统计建模的多极化SAR图像分类方法。文中研究了基于copulas函数多极化SAR图像统计建模方法,并借鉴字典集的思想提出了一种基于字典集的混合copulas建模方法。实验证实,对于多极化SAR图像统计分布建模,该方法有更好的数据描述能力。同时也验证了,与单极化SAR图像分类相比较,多极化SAR图像分类由于综合利用了多通道数据信息,能够取得更好的分类结果。
As a tool of acquiring information, SAR is of significant strategic sense, the interpretation of SAR imagery is a very important task. Terrain classification plays an important part in the interpretation of SAR imagery. The classification method based on MRF model and statistic modeling, which is able to unify the contextual information from label images and statistical properties of observed images, was widely used and developed in image processing. In this thesis, SAR image classification method based on MRF and statistical modeling is thoroughly studied.
Firstly, as the foundation theory, SAR clutter statistical model is studied in depth. According to existing work and the progresses made recently, the relevant techniques of statistical modeling of clutter in SAR image are comprehensively reviewed, and a parameter estimation approach stemming from Mellin transform which is also named“Method of Log-Cumulant”(MoLC) is introduced and analyzed in detail. Based on MoLC, the estimate equations of the statistical distribution models mentioned are induced and comprehensively summarized. To satisfy the requirement in the modeling of high-resolution SAR imagery, a recently proposed method named“Dictionary based stochastic expectation maximization for SAR amplitude probability density function”is studied. Copula, a statistical tool that was designed for constructing joint distributions from marginals with a wide variety of allowable dependence structures, is introduced to model the join distribution of multi-channel space of data.
Secondly, the classification of single polarized SAR imagery based on MRF model and statistical modeling is studied. Based on MRF model, the process of image classification is turned to be a problem of energy-minimum by seeking the optimal label Matrix. A classification method based on MRF model and the joint distribution of the data space of gray level and texture information (Contrast Statistics) deduced from Gray Level Co-occurrence Matrix (GLCM) via copulas is proposed.
Finally, the classification of multi-polarized SAR imagery based on MRF model and statistical modeling is studied. Statistical modeling approach based on copula theory is studied. After the in depth study of finite mixture copulas model, a dictionary based mixture copulas model is proposed, which has good statistical representation ability. Experiments verified the better performance, compared with the classification process only using single-polarized data.
引文
[1]付琨.高分辨率单视单极化SAR图像地物分类方法研究[D].长沙:国防科学技术大学, 2002.
[2]倪心强. SAR图像分类与自动目标识别技术研究[D].北京:中国科学院, 2007.
[3]高贵. SAR图像目标ROI自动获取技术研究[D].长沙:国防科学技术大学, 2007.
[4]周晓光.极化SAR图像分类方法研究[D].长沙:国防科学技术大学, 2008.
[5] L. M. Novak,et al. The Automatic Target-Recognition System in SAIP. The Lincoln Laboratory Journal,1997,10(2):187-202.
[6] R. Dressler,et al. COTS SAR Processing Software. IEEE National RadarConference, 1996: 136-141.
[7] M. Greenspan, et al. Development and Evaluation of A Real Time SAR ATR System. IEEE, 1998, 38-43.
[8] http://www.atl.external.lmco.com/projects/rassp/RASSP_legacy/casestudies/ SAR/CAS-SAR_02.HTM
[9] http://www.alphatech.com/secondary/techpro/projects/mstar/MSTAR_Top Level.html
[10] R.A. English, et al. Development of an ATR Workbench for SAR Imagery.Technical Report, DRDC, Ottawa, 2005.
[11] J.C. Oliver, et al. http://www.infosar.co.uk/misc/demo.html.
[12] W. Roller, et al. Detection and Recognition of Vehicles in High-Resolution SAR Imagery. SPIE, 2001, 4380:142-152.
[13]汪伟.基于偏微分方程的人工地物与自然区域分类技术研究[D].上海:上海交通大学, 2007.
[14] Haralick RM, Shanmugam K, Dinstein I. Textural features for image classification[J]. IEEE Transon Systems, Mans and Cybern, 1973,3(6): 610-621.
[15]安国成,刘振华,于文震.基于分形特征的高分辨率SAR图像分类[J].现代雷达, 2006,28(6): 26-29.
[16] Kenneth.Falconer.分形几何-数学基础及其应用[M], 2003.
[17] S.Ranjan U, Narayana A. Classi?cation of objects in SAR images using scaling features[J].
[18]孙即祥.图像处理[M].北京:科学出版社, 2004.
[19] Mallat SG. A theory for multiresolution signal decomposition: the wavelet representation. Pattern Analysis and Machine Intelligence, IEEE Transactions on[C], 1989:674-693.
[20]张斌,高鑫.一种基于Gabor滤波器组和LBP的多特征联合SAR纹理图像分类方法[J].科学技术与工程, 2010,(17): 4196-4201.
[21]韩萍,吴仁彪,王兆华.基于KPCA准则的SAR目标特征提取与识别[J].电子与信息学报, 2003,25(10): 1297-1301.
[22]胡风明,王满玉,范学花.基于Gabor滤波器和ICA技术的SAR目标识别算法[J].现代雷达, 2011,(06): 32-36.
[23]孙即祥.现代模式识别[M].北京:高等教育出版社, 2008.
[24] Macchiavello G, Moser G, Boni G, et al. Automatic unsupervised classification of snow-covered areas by decision-tree classification and minimum-error thresholding. Geoscience and Remote Sensing Symposium,2009 IEEE International,IGARSS 2009[C], 2009:1000-1003.
[25] Geman S, Geman D. Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1984,6(6): 721-741.
[26]丁明涛. SAR图像分类的双Markov随机场模型研究[D].西安:西北工业大学, 2007.
[27]黄宁,朱敏慧,张守融.一种采用高斯隐马尔可夫随机场模型的遥感图像分类算法[J].电子与信息学报, 2003,(01): 50-53.
[28]周晓光,贺志国,匡纲要,等.基于极化G0分布和MRF的多视PolSAR图像迭代分类方法[J].宇航学报, 2009,(01): 276-281.
[29]郑玮,康戈文,陈武凡,等.基于模糊马尔可夫随机场的无监督遥感图像分割算法[J].遥感学报, 2008,(02): 246-252.
[30]张强,吴艳.基于上下文和隐类属的小波域马尔可夫随机场SAR图像分割[J].电子与信息学报, 2008,(01): 211-215.
[31]侯一民,郭雷.一种基于马尔可夫随机场的SAR图像分割新方法[J].电子与信息学报, 2007,(05): 1069-1072.
[32]夏桂松,何楚,孙洪.一种基于非参数密度估计和马尔可夫上下文的SAR图像分割算法[J].电子与信息学报, 2006,28(12): 2209-2213.
[33] Tran TN, Wehrens R, Hoekman DH, et al. Initialization of Markov random field clustering of large remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005,43: 1912 -1919.
[34]郦苏丹张王.基于马尔可夫随机场的SAR图象目标分割[J].中国图象图形学报, 2002,(08): 794-799.
[35] Dubes RC, Jain AK. Random field models in image analysis[J]. Journal of Applied Statistics, 1989,16(2): 131-164.
[36] Wu Y, Ji K, Yu W, et al. Region-Based Classification of Polarimetric SARImages Using Wishart MRF[J]. IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2008,5(4): 668-672.
[37]刘梦玲,何楚,苏鑫,等.基于pLSA和Topo-MRF模型的SAR图像分类算法研究[J].武汉大学学报(信息科学版), 2011,(01): 122-125.
[38]刘爱平,付琨,尤红建,等.基于MAR-MRF的SAR图像分割方法[J].电子与信息学报, 2009,(11): 2557-2562.
[39]焦李成.神经网络计算[M].西安:西安电子科技大学出版社, 1993.
[40]殷勤业.模式识别与神经网络[M].北京:机械工业出版社, 1992.
[41]李海权,李春霞,吴彩银,等.基于纹理和BP神经网络的SAR图像分类[J].遥感信息, 2009,(3): 58-63.
[42]段连飞,黄国满,荣伟,等.基于BP神经网络的机载高分辨率SAR图像分类方法研究[J].测绘通报, 2009,(2): 14-17,27.
[43]梁小祎.基于纹理分析的溢油SAR图像分类研究[D].大连:大连海事大学, 2007.
[44]李晓辉,谢建红,高鑫.基于Brushlet和RBF网络的SAR图像分类[J].微计算机信息, 2009,(6): 295-297.
[45] Xu J, Lu L, Feng Z, et al. SAR image feature extraction and classification with fractal-based description. Signal Processing, 2002 6th International Conference on[C], 2002:1392-1395.
[46]周志华,陈兆乾,陈世福.自适应谐振理论综述[J].计算机科学, 1999,(04): 54-79.
[47]龚淑蕾,张煜东,沈士喆,等.一种极化合成孔径雷达图像分类的混合方法[J].东南大学学报(自然科学版), 2009,(S1): 294-298.
[48]王瑞花,宋建社.基于改进FCM算法的SAR图像分类[J].西北大学学报:自然科学版, 2008,38(4): 574-578.
[49] Bezdek JC. Pattern Recognition with Fuzzy Objective Function Algorithms. New York: Plenum Press, 1981:22-56.
[50]谈璐璐,张涛,杨汝良.基于模糊聚类的PolInSAR数据非监督分类方法[J].系统工程与电子技术, 2011,(02): 305-309.
[51]曾亮,陈华杰,林岳松.基于模糊松弛的极化图像分类算法[J].杭州电子科技大学学报, 2010,(04): 169-172.
[52]马文萍,焦李成,张向荣,等.基于量子克隆优化的SAR图像分类[J].电子学报, 2007,35(12): 2241-2246.
[53]张向荣,谭山,焦李成.基于商空间粒度计算的SAR图像分类[J].计算机学报, 2007,30(3): 483-490.
[54]吴永辉,计科峰,郁文贤.利用SVM的全极化、双极化与单极化SAR图像分类性能的比较[J].遥感学报, 2008,(01): 46-52.
[55]王红艳.基于支持向量机的合成孔径雷达图像分割[D].西安:西安科技大学, 2010.
[56]汤井田,胡丹,龚智敏.基于SVM的SAR图像分类研究[J].遥感技术与应用, 2008,23(3): 341-345.
[57]吴永辉,计科峰,李禹,等.利用SVM的极化SAR图像特征选择与分类[J].电子与信息学报, 2008,(10): 2348-2351.
[58]吴永辉,计科峰,郁文贤.基于支持向量机的极化SAR图像分类[J].现代雷达, 2007,(06): 57-60.
[59]李长春,冒亚明,孙灏,等.基于多通道分类合成的SAR图像分类研究[J].计算机与现代化, 2010,(3): 7-11.
[60]员永生.基于支持向量机分类的面向对象土地覆被图像分类方法研究[D].杨凌:西北农林科技大学, 2010.
[61] Moser G, Serpico SB. Contextual remote-sensing image classification by support vector machines and Markov random fields. Geoscience and Remote Sensing Symposium (IGARSS), 2010 IEEE International[C], 2010:3728-3731.
[62] Tusia D. Multisource Composite Kernels for Urban-Image Classification[J]. IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2010,7(1): 88-92.
[63] Yuan C, Casasent D. A new SVM for distorted SAR object classification. SPIE[C], 2005:10-22.
[64]刘若辰,邹海双,张莉,等.一种改进的Laplacian SVM的SAR图像分割算法[J].红外与毫米波学报, 2011,(03): 250-254,259.
[65]张中山,燕琴,余洁,等.基于粒子群算法的全极化SAR图像非监督分类算法研究[J].武汉大学学报(信息科学版), 2010,(08): 941-945.
[66] Lee J-S, Grunes MR, Pottier E. Quantitative comparison of classification capability: fully polarimetric versus dual and single-polarization SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2001,39: 2343 -2351.
[67] Nicolas J-m. Introduction to Second kind statistics:Application to SAR Image Laws(in French)[M]. Paris, 2002.
[68] Nelsen RB. An introduction to copulas[M]. New York: Springer, 2007.
[69] Oliver C, Quegan S. Understanding Synthetic Aperture Radar Images[M]. Raleigh: SciTech Publishing, 2004.
[70]匡纲要,高贵,蒋咏梅.合成孔径雷达目标检测理论、算法及应用[M].长沙:国防科技大学出版社, 2007.
[71]张琦.基于统计模型的SAR图像车辆目标检测方法研究[D].长沙:国防科学技术大学, 2005.
[72]赵晶晶.红外图像中低信噪比小目标处理方法研究[D].长沙:国防科学技术大学, 2007.
[73]邹焕新. SAR图像舰船目标与航迹检测方法研究[D].长沙:国防科技大学, 2003.
[74] E.Parzen. On estimation of probability density function and mode[J]. Sigal Processing, 1962,33: 267-281.
[75] C.M.Bishop. Neural Networks for Pattern Recognition[M]: Oxford University Press, 1996.
[76] Moser G, Zerubia J, Serpico SB. SAR amplitude probability density function estimation based on a generalized Gaussian model[J]. IEEE Transactions on Image Processing, 2006,15: 1429-1442.
[77] Li H-C, Hong W, Wu Y-R, et al. On the Empirical-Statistical Modeling of SAR Image[J]. IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING, 2011,5(3): 386-397.
[78] Li H-C, Hong W, Wu Y-R. An efficient and flexible statistical model based on generalized gamma distribution for SAR amplitude[J]. IEEE Transactions on Geoscience and Remote Sensing, 2010,48(6): 2711-2722.
[79]盛骤,谢式千,潘承毅.概率论与数理统计[M].北京:高等教育出版社, 1997.
[80] Salazar JS. Detection Schemes for Synthetic Aperture Radar Inagery Based On a Beta Prime Statistical Model[D]: Florida Univesity, 1999.
[81] Delignon Y. Modeling Non-rayleigh Speckle Distribution in SAR Images[J]. IEEE Trans Geosci Remote Sens, 2002,40(6): 1430-1435.
[82]高贵. SAR图像统计建模研究综述[J].信号处理, 2009,(08): 1270-1278.
[83] Kuruoglu EE, Zerubia J. Modeling SAR images with a generalization of the Rayleigh distribution[J]. IEEE Transactions on Image Processing, 2004,13: 527-533.
[84]孙增国,韩崇昭.拖尾Rayleigh分布:基本性质及其应用[J].自动化学报, 2008,(09): 1067-1075.
[85]孙增国,韩崇昭.基于负数阶矩的正值Alpha稳定分布的参数估计[J].西安交通大学学报, 2007,(06): 645-649.
[86] Krylov V, Moser G, Serpico SB, et al. On the Method of Logarithmic Cumulants for Parametric Probability Density Function Estimation. 2011.
[87] Krylov VA, Zerubia J. Generalized gamma mixtures for supervised SAR image classification[J]. 2010:
[88] K.Song. Globally convergent algorithms for estimating generalized gammadistribution in fast signal and image processing[J]. IEEE Transactions on Image Processing, 2008,17(8): 1233-1250.
[89] Frery AC, Muller H-J, Yanasse CCF, et al. A model for extremely heterogeneous clutter[J]. IEEE Transactions on Geoscience and Remote Sensing, 1997,35(3): 648-659.
[90]时公涛,高贵,周晓光,等.基于Mellin变换的G0分布参数估计方法[J].自然科学进展, 2009,19(6): 677-690.
[91] Tison C, Nicolas J-M, Tupin F. Classification of X-band high resolution SAR images over urban areas: Markovian segmentation using Mellin transform. Remote Sensing and Data Fusion over Urban Areas, 2003 2nd GRSS/ISPRS Joint Workshop on[C], 2003:110-114.
[92] Moser G, Zerubia J, Serpico SB. Dictionary-based stochastic expectation-maximization for SAR amplitude probability density function. 2004.
[93] Nicolas JM, Tupin F. Gamma mixture modeled with "second kind statistics": application to SAR image processing. IGARSS[C], 2002:2489 -2491.
[94]时公涛.基于干涉图的多通道SAR地面慢动目标自动检测技术研究[D].长沙:国防科学技术大学, 2009.
[95] Tison C, Nicolas J-M, Tupin F, et al. A new statistical model for Markovian classification of urban areas in high-resolution SAR images[J]. IEEE Transactions on Geoscience and Remote Sensoring, 2004,42(10): 2046-2057.
[96]孙增国,韩崇昭.基于拖尾分布的高分辨率合成孔径雷达图像建模[J].物理学报, 2010,(2): 998-1008.
[97] Nicolas J-M, A.Maruani. Lower-order statistics:a new approach for probability density functions defined on R+. EUSIPCO[C]. Tampere,Finland, 2000.
[98]皇甫堪,陈建文,楼生强.现代数字信号处理[M].北京:电子工业出版社, 2004.
[99] Figueiredo MAT, Jain AK. Unsupervised learning of finite mixture models[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24: 381-396.
[100] Delignon Y, Marzouki A, Pieczynski W. Estimation of generalized mixtures and its application in image segmentation[J]. Image Processing, IEEE Transactions on, 1997,6: 1364-1375.
[101] Biernacki C, Celeux G. Strategies for getting highest likelihood in mixture models. INRIA, 2001.
[102] CELEUX G, CHAUVEAU D, DIEBOLT J. On Stochastic Versions of the EM Algorithm. 1995.
[103] celeux G, DIEBOLT J. The EM and SEM Algorithms for mixtures:Statisticals and numerical aspect. INRIA, 1987.
[104] Biernacki C, Celeux G. Choosing starting values for the EM algorithm for getting the highest likelihood in multivariate Gaussian mixture models.
[105] Krylov V, Moser G, Serpico SB, et al. Dictionary-based probability density function estimation for high-resolution SAR data. Procedings of SPIE-IS&T Electronic Imaging[C], 2009.
[106] Krylov VA, Moser G, Serpico SB, et al. Enhanced Dictionary-Based SAR Amplitude Distribution Estimation and Its Validation With Very High-Resolution Data[J]. IEEE Geoscience and Remote Sensing Letters, 2011,8(1): 148-152.
[107] Krylov V, Moser G, Serpico SB, et al. Modeling the statistics of high resolution SAR images. INRIA, 2009.
[108] Sklar A. Fonctions de repartitionàn dimensions et leurs marges[M]. Paris: Publication de 1'Instit de Statistique de 1'Université, 1959.
[109]刘琼芳.基于Copula理论的金融序列相依性研究[D].重庆:重庆大学, 2010.
[110]葛镇忠.基于随机森林和Copula的港口物流能力研究[D].天津:天津大学, 2010.
[111] Krylov V, Moser G, Serpico SB, et al. Multichannel SAR image classification by finite mixtures, copula theory and Markov random fields. Proceedings of AIP[C]. Chamonix, France, 2010:299-306.
[112] Krylov V, Moser G, Serpico SB, et al. Supervised High Resolution Dual Polarization SAR Image Classification by Finite Mixtures and Copulas[J]. IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING, 2011:
[113] Moser G, Krylov V, Serpico SB, et al. High resolution SAR-image classification by Markov random fields and finite mixtures. IS&TSPIE Electronic Imaging[C], 2010.
[114] Voisin A, Moser G, Krylov V, et al. Classification of very high resolution SAR images of urban areas by dictionary-based mixture models, copulas and Markov random fields using textural features. Proceedings of SPIE[C]. Toulouse, France, 2010:78300O-78301--78300O-78311.
[115]黄雁勇.混合Copula的选择_估计及其应用[D].西安:西安交通大学, 2007.
[116]王润生.图像理解[M].长沙:国防科技大学出版社, 1998.
[117] Kottk DP, Fiore PD, Brown KL, et al. Design for HMM-based SAR ATR. SPIE[C]. Orland, Florida, 1998:541-551.
[118] Kato Z, Zerubia J, Berthod M. Satellite image classification using a modified Metropolis dynamics. IEEE International Conference on Acoustics, Speech, and Signal Processing [C], 1992:573-576.
[119] Foody GM. status of landcover classification accuracy assessment[J]. 1992,58(10): 1459-1460.
[120]付信际.合成孔径雷达图像分类与目标检测技术研究[D].北京:中国科学院, 2005.
[121] Krylov VA, Moser G, Serpico SB, et al. Supervised High-Resolution Dual-Polarization SAR Image Classification by Finite Mixtures and Copulas[J]. IEEE J Sel Top Signal Process, 2011,5(5): 554--566.
[122] Kojadinovic I, Yan J. Comparison of three semiparametric methods for estimating dependence parameters in copula models[J]. Insurance: Mathematics and Economics, 2010,47(1): 52-63.
[123]吴永辉,计科峰,郁文贤. SVM全极化SAR图像分类中的特征选择[J].信号处理, 2007,23(6): 877-881.
[124]王海江.极化SAR图像分类方法研究[D].成都:电子科技大学, 2008.
[125] Lee J-S, Hoppel KW, Mango SA, et al. Intensity and phase statistics of multilook polarimetric and interferometric SAR imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 1994,32: 1017 -1028.
[126]吴永辉.极化SAR图像分类技术研究[D].长沙:国防科学技术大学, 2007.
[127] Freitas CC, Frery AC, COrreia AH. The polarimetric G distribution for SAR data analysis[J]. Environmentrics, 2005,16(1): 13-31.
[128] Frery AC, Correia AH, Freitas CC. Multifrequency full polarimetric SAR classification with multiple sources of statistical evidence. International Geoscience and Remote Sensing Symposium[C]. Denver, Colorado, USA, 2006:4195-4197.