Characterization of SAR Clutter and Its Applications to Land and Ocean Observations

Preface

Contents

1 Overview for Statistical Modeling of SAR Images

1.1 Introduction

1.2 Model Classification and Research Contents

1.2.1 Parameter Estimation

1.2.2 Goodness-of-Fit Tests

1.3 Statistical Models

1.3.1 Nonparametric Models

1.3.2 Parametric Models

1.4 Classification of Parametric Models

1.4.1 The Statistical Models Developed from the Product Model

1.4.2 The Statistical Model Developed from the Generalized Central Limit Theorem

1.4.3 The Empirical Distributions

1.4.4 Other Models

1.5 The Relationship Among the Major Models and Their Applications

1.5.1 The Relationship Among the Parametric Statistical Models

1.5.2 Summary of the Applications of the Major Models

1.6 Discussion of Future Work

1.7 Conclusions

References

2 Statistical Modeling of Single-Channel SAR Images

2.1 Modeling SAR Images Based on a Generalized Gamma Distribution for Texture Component

2.1.1 The Proposed G?? Model

2.1.2 Parameter Estimator of the G ? ? Model Based on MoLC

2.1.3 Experimental Results

2.1.4 Appendix 2-A. The Derivation of mth Order Moments of the G ? ? Distribution

2.1.5 Appendix 2-B. Proof of the Relationship Between Distributions

2.2 Scheme for Characterizing Clutter Statistics in SAR Amplitude Images by Combining Two Parametric Models

2.2.1 mathcalG_AO Model

2.2.2 Parameter Estimates of the G?D

2.2.3 Analytical Conditions of Applicability

2.2.4 Proposed Scheme

2.2.5 Experimental Results and Analysis

2.3 An Improved Scheme for Parameter Estimation of G0 Distribution Model in High-Resolution SAR Images

2.3.1 The G0 Model

2.3.2 MoM Based Parameter Estimation

2.3.3 MT Based Parameter Estimation

2.3.4 Our Proposed Parameter Estimation

2.3.5 Results and Discussion

2.4 Conclusions

References

3 Target Detection and Terrain Classification of Single-Channel SAR Images

3.1 A CFAR Detection Algorithm for Generalized Gamma Distributed Background in High-Resolution SAR Images

3.1.1 Generalized Gamma Distribution and Its Estimation

3.1.2 CFAR Algorithm Using G ?D for Background

3.1.3 Performance Evaluation

3.2 A Parzen Window Kernel Based CFAR Algorithm for Ship Detection in SAR Images

3.2.1 Statistical Modeling of SAR Image Based on Parzen Window Kernel

3.2.2 CFAR Detection

3.2.3 Experimental Results

3.3 A Markovian Classification Method for Urban Areas in High-Resolution SAR Images

3.3.1 Markovian Formalism

3.3.2 Optimization Algorithm

3.3.3 Results and Analysis

3.4 Conclusion

References

4 Statistical Modeling of Multi-channel SAR Images

4.1 Introduction

4.2 Normalized Interferogram

4.3 The Joint Distribution

4.3.1 The Known Joint Distribution for Heterogeneous Regions

4.4 The Proposed Distribution for Interferogram’s Magnitude of Homogenous Clutter

4.4.1 The ?_mathcalIn Distribution for Homogeneous Clutter

4.4.2 Parameter Estimators of  ?_mathcalIn

4.5 Statistics of Multilook SAR Interferogram for In-homogeneous Clutter Based on  ?_mathcalIn

4.5.1 Extremely Heterogeneous Clutter

4.5.2 Heterogeneous Clutter

4.5.3 Parameter Estimators of In-homogeneous Clutter Statistics

4.5.4 Relationship Between Distributions

4.5.5 Experimental Analysis

Appendix 4.1

References

5 Moving Vehicle Detection in Along-Track Interferometric SAR Complex Images

5.1 Introduction

5.2 The IMP Metric

5.2.1 The Characteristics of Moving Targets Compared to Stationary Clutter

5.2.2 The Construction of the New Detection Metric

5.3 Statistical Distribution Model of IMP Metric

5.3.1 Homogeneous Area

5.3.2 The mathcalS^0 Distribution

5.3.3 The Parametric Estimators of the  mathcalS^0 Distribution

5.4 CFAR Detection

5.4.1 The Threshold Derivation

5.4.2 Detailed Flow of CFAR Detection

5.4.3 Experimental Results

5.5 Conclusion

Appendix 5.1: The Derivation of the  mathcalS^0 Distribution

Appendix 5.2: The Second-Kind First Characteristic Function of the  mathcalS^0 Distribution

References

6 Statistical Modeling and Target Detection of PolSAR Images

6.1 Introduction

6.2 Multiplicative Model for Covariance Matrix

6.2.1 Multilook PolSAR Data

6.2.2 Multiplicative Model

6.3 Statistical Modeling of PolSAR Images with Generalized Gamma Distribution for Backscatter

6.3.1 Advantage of G ?D

6.3.2 The Compound Model

6.3.3 Estimator Based on Method of Matrix Log-Cumulants

6.4 Experimental Results and Discussions

6.4.1 Experimental Data and Evaluation Criteria

6.4.2 Modeling Result

6.4.3 Discussions

6.5 Ship Detection in High-Resolution Dual Polarization SAR Amplitude Images

6.5.1 Dual-Pol SAR Data Description

6.5.2 The PMA Detector

6.5.3 The CFAR Algorithm of PMA Detector

6.5.4 Experimental Results and Analysis

6.5.5 Experimental Results and Analysis

Appendix 6.1: The Derivation of  CG?_P Distribution Toward the  mathcalK_P and  mathcalG_P^0 Distributions

Appendix 6.2: The Derivation of the Distribution of  B_1 B_2

Appendix 6.3: The Approximate PDF for  ?

References