Characterization of SAR Clutter and Its Applications to Land and Ocean Observations (eBook)

Gui Gao received a B.S. in Information Engineering, and M.S. and Ph.D. degrees in remote sensing information processing from the National University of Defense Technology (NUDT), Changsha, China, in 2002, 2003 and 2007, respectively. In 2007, he joined the Faculty of Information Engineering, School of Electronic Science and Engineering, NUDT, where he is currently an Associate Professor. Since 2016, he has been a Distinguished Professor at the School of Information Science and Engineering, Central South University, Changsha, China. He has authored over 80 journal papers and written three books. His current research interests include SAR ATR (automatic target recognition), statistical modeling of SAR images, SAR ship detection, and SAR GMTI (ground moving target indication). Dr. Gao is a member of the IEEE Geoscience and Remote Sensing Society, the Applied Computational Electromagnetics Society, and the Chinese Institute of Electronics (CIE), and a member of the Young Scientist Forum of CIE. He received the Hunan Province Excellent Master’s Thesis Award in 2006, the Chinese Army Excellent Doctoral Thesis Award in 2008, Outstanding Young People in NUDT and Hunan Province Award in 2014 and 2016, and the Natural Science in Hunan Province Award. He was also selected as one of the Young Talents of Hunan in 2016. He was the Lead Guest Editor of the International Journal of Antenna and Propagation. He is on the editorial board of the Chinese Journal of Radars. He has also been the Co-Chairman of several conferences in the field of remote sensing. He was the Excellent Reviewer for the journal of Xi’an Jiaotong University in 2013.  

Preface 5
Contents 7
1 Overview for Statistical Modeling of SAR Images 11
1.1 Introduction 11
1.2 Model Classification and Research Contents 12
1.2.1 Parameter Estimation 13
1.2.2 Goodness-of-Fit Tests 13
1.3 Statistical Models 14
1.3.1 Nonparametric Models 14
1.3.2 Parametric Models 15
1.4 Classification of Parametric Models 15
1.4.1 The Statistical Models Developed from the Product Model 16
1.4.2 The Statistical Model Developed from the Generalized Central Limit Theorem 21
1.4.3 The Empirical Distributions 21
1.4.4 Other Models 22
1.5 The Relationship Among the Major Models and Their Applications 23
1.5.1 The Relationship Among the Parametric Statistical Models 23
1.5.2 Summary of the Applications of the Major Models 24
1.6 Discussion of Future Work 24
1.7 Conclusions 28
References 28
2 Statistical Modeling of Single-Channel SAR Images 33
2.1 Modeling SAR Images Based on a Generalized Gamma Distribution for Texture Component 33
2.1.1 The Proposed G?? Model 34
2.1.2 Parameter Estimator of the G ? ? Model Based on MoLC 36
2.1.3 Experimental Results 39
2.1.4 Appendix 2-A. The Derivation of mth Order Moments of the G ? ? Distribution 43
2.1.5 Appendix 2-B. Proof of the Relationship Between Distributions 45
2.2 Scheme for Characterizing Clutter Statistics in SAR Amplitude Images by Combining Two Parametric Models 46
2.2.1 mathcalG_AO Model 47
2.2.2 Parameter Estimates of the G?D 48
2.2.3 Analytical Conditions of Applicability 48
2.2.4 Proposed Scheme 52
2.2.5 Experimental Results and Analysis 54
2.3 An Improved Scheme for Parameter Estimation of G0 Distribution Model in High-Resolution SAR Images 63
2.3.1 The G0 Model 63
2.3.2 MoM Based Parameter Estimation 66
2.3.3 MT Based Parameter Estimation 67
2.3.4 Our Proposed Parameter Estimation 69
2.3.5 Results and Discussion 70
2.4 Conclusions 80
References 82
3 Target Detection and Terrain Classification of Single-Channel SAR Images 84
3.1 A CFAR Detection Algorithm for Generalized Gamma Distributed Background in High-Resolution SAR Images 84
3.1.1 Generalized Gamma Distribution and Its Estimation 85
3.1.2 CFAR Algorithm Using G ?D for Background 86
3.1.3 Performance Evaluation 88
3.2 A Parzen Window Kernel Based CFAR Algorithm for Ship Detection in SAR Images 93
3.2.1 Statistical Modeling of SAR Image Based on Parzen Window Kernel 94
3.2.2 CFAR Detection 95
3.2.3 Experimental Results 97
3.3 A Markovian Classification Method for Urban Areas in High-Resolution SAR Images 102
3.3.1 Markovian Formalism 103
3.3.2 Optimization Algorithm 104
3.3.3 Results and Analysis 104
3.4 Conclusion 108
References 109
4 Statistical Modeling of Multi-channel SAR Images 111
4.1 Introduction 111
4.2 Normalized Interferogram 112
4.3 The Joint Distribution 114
4.3.1 The Known Joint Distribution for Heterogeneous Regions 114
4.4 The Proposed Distribution for Interferogram’s Magnitude of Homogenous Clutter 115
4.4.1 The ?_mathcalIn Distribution for Homogeneous Clutter 115
4.4.2 Parameter Estimators of  ?_mathcalIn 118
4.5 Statistics of Multilook SAR Interferogram for In-homogeneous Clutter Based on  ?_mathcalIn 120
4.5.1 Extremely Heterogeneous Clutter 120
4.5.2 Heterogeneous Clutter 121
4.5.3 Parameter Estimators of In-homogeneous Clutter Statistics 121
4.5.4 Relationship Between Distributions 123
4.5.5 Experimental Analysis 124
Appendix 4.1 128
References 129
5 Moving Vehicle Detection in Along-Track Interferometric SAR Complex Images 131
5.1 Introduction 131
5.2 The IMP Metric 132
5.2.1 The Characteristics of Moving Targets Compared to Stationary Clutter 132
5.2.2 The Construction of the New Detection Metric 132
5.3 Statistical Distribution Model of IMP Metric 134
5.3.1 Homogeneous Area 134
5.3.2 The mathcalS^0 Distribution 135
5.3.3 The Parametric Estimators of the  mathcalS^0 Distribution 137
5.4 CFAR Detection 137
5.4.1 The Threshold Derivation 137
5.4.2 Detailed Flow of CFAR Detection 138
5.4.3 Experimental Results 139
5.5 Conclusion 142
Appendix 5.1: The Derivation of the  mathcalS^0 Distribution 142
Appendix 5.2: The Second-Kind First Characteristic Function of the  mathcalS^0 Distribution 143
References 143
6 Statistical Modeling and Target Detection of PolSAR Images 145
6.1 Introduction 145
6.2 Multiplicative Model for Covariance Matrix 146
6.2.1 Multilook PolSAR Data 146
6.2.2 Multiplicative Model 147
6.3 Statistical Modeling of PolSAR Images with Generalized Gamma Distribution for Backscatter 148
6.3.1 Advantage of G ?D 148
6.3.2 The Compound Model 150
6.3.3 Estimator Based on Method of Matrix Log-Cumulants 151
6.4 Experimental Results and Discussions 155
6.4.1 Experimental Data and Evaluation Criteria 155
6.4.2 Modeling Result 156
6.4.3 Discussions 159
6.5 Ship Detection in High-Resolution Dual Polarization SAR Amplitude Images 160
6.5.1 Dual-Pol SAR Data Description 161
6.5.2 The PMA Detector 163
6.5.3 The CFAR Algorithm of PMA Detector 165
6.5.4 Experimental Results and Analysis 167
6.5.5 Experimental Results and Analysis 168
Appendix 6.1: The Derivation of  CG?_P Distribution Toward the  mathcalK_P and  mathcalG_P^0 Distributions 169
Appendix 6.2: The Derivation of the Distribution of  B_1 B_2 171
Appendix 6.3: The Approximate PDF for  ? 171
References 172