Hyperspectral Image Processing (eBook)

Preface 5
Contents 8
Symbols and Abbreviations 14
Abstract 16
1 Basic Theory and Main Processing Techniques of Hyperspectral Remote Sensing 17
1.1 Basic Theory of Hyperspectral Remote Sensing 17
1.1.1 Theory of Remote Electromagnetic Wave 17
1.1.2 Interaction of Solar Radiation and Materials 18
1.1.3 Imaging Spectrometer and Spectral Imaging Modes 19
1.1.4 Imaging Characteristics of HSI 23
1.2 Classification Technique of HSI 24
1.2.1 Supervised Classifications and Unsupervised Classifications 24
1.2.2 Parameter Classifications and Nonparameter Classifications 27
1.2.3 Crisp Classifications and Fuzzy Classifications 29
1.2.3.1 Crisp Classifications 29
1.2.4 Other Classification Methods 29
1.3 Endmember Extraction Technique of HSI 30
1.4 Spectral Unmixing Technique of HSI 33
1.4.1 Nonlinear Model 34
1.4.2 Linear Model 35
1.4.3 Multi-endmember Mode of Linear Model 39
1.5 Sub-pixel Mapping Technique of HSI 40
1.5.1 Spatial Correlation-Based Sub-pixel Mapping 42
1.5.2 Spatial Geostatistics-Based Sub-pixel Mapping 44
1.5.3 Neural Network-Based Sub-pixel Mapping 45
1.5.4 Pixel-Swapping Strategy-Based Sub-pixel Mapping 46
1.6 Super Resolution Technique of HSI 48
1.7 Anomaly Detection Technique of HSI 51
1.8 Dimensionality Reduction and Compression Technique for HSI 54
1.8.1 Dimensionality Reduction: Band Selection and Feature Extraction 54
1.8.2 Compression: Lossy Compression and Lossless Compression 58
References 60
2 Classification Technique for HSI 61
2.1 Typical Classification Methods 61
2.2 Typical Assessment Criterions 64
2.3 SVM-Based Classification Method 66
2.3.1 Theory Foundation 66
2.3.2 Classification Principle 68
2.3.3 Construction of Multi-class Classifier with the Simplest Structure 76
2.3.4 Least Squares SVM and Its SMO Optimization Algorithm 79
2.3.5 Triply Weighted Classification Method 82
2.4 Performance Assessment for SVM-Based Classification 86
2.4.1 Performance Assessment for Original SVM-Based Classification 88
2.4.2 Performance Assessment for Multi-class Classifier with the Simplest Structure 89
2.4.3 Performance Assessment for Triply Weighted Classification 90
2.5 Chapter Conclusions 92
References 93
3 Endmember Extraction Technique of HSI 94
3.1 Endmember Extraction Method: N-FINDR 94
3.1.1 Introduction of Related Theory 94
3.1.2 N-FINDR Algorithm 97
3.2 Distance Measure-Based Fast N-FINDR Algorithm 99
3.2.1 Substituting Distance Measure for Volume One 99
3.2.2 PPI Concept-Based Pixel Indexing 101
3.2.3 Complexity Analysis and Efficiency Assessment 102
3.3 Linear LSSVM-Based Distance Calculation 102
3.4 Robust Method in Endmember Extraction 104
3.4.1 In the Pre-processing Stage: Obtaining of Robust Covariance Matrix 104
3.4.2 In Endmember Extraction Stage: Deletion of Outliers 107
3.5 Performance Assessment 107
3.5.1 Distance Measure-Based N-FINDR Fast Algorithm 107
3.5.2 Robustness Assessment 109
3.6 Two Applications of Fast N-FINDR Algorithm 113
3.6.1 Construction of New Solving Algorithm for LSMM 113
3.6.2 Construction of Fast and Unsupervised Band Selection Algorithm 114
3.7 Chapter Conclusions 118
References 118
4 Spectral Unmixing Technique of HSI 120
4.1 LSMM-Based LSMA Method 120
4.2 Two New Solving Methods for Full Constrained LSMA 123
4.2.1 Parameter Substitution Method in Iteration Solving Method 123
4.2.2 Geometric Solving Method 124
4.3 The Principle of LSVM-Based Spectral Unmixing 129
4.3.1 Equality Proof of LSVM and LSMM for Spectral Unmixing 129
4.3.2 The Unique Superiority of LSVM-Based Unmixing 131
4.4 Spatial--Spectral Information-Based Unmixing Method 132
4.5 SVM-Based Spectral Unmixing Model with Unmixing Residue Constraints 133
4.5.1 Original LSSVM-Based Spectral Unmixing 134
4.5.2 Construction of Spectral Unmixing Model Based on Unmixing Residue Constrained LSSVM and Derivation of Its Closed form Solution 136
4.5.3 Substituting Multiple Endmembers for Single One in the New Model 139
4.6 Performance Assessment 140
4.6.1 Performance Assessment for Original SVM-Based Spectral Unmixing 140
4.6.2 Assessment on Robust Weighted SVM-Based Unmixing 142
4.6.3 Assessment on Spatial--Spectral Unmixing Method 144
4.6.4 Performance Assessment on New SVM Unmixing Model with Unmixing Residue Constraints 146
4.7 Fuzzy Method of Accuracy Assessment of Spectral Unmixing 150
4.7.1 Fuzzy Method of Accuracy Assessment 150
4.7.2 Application of Fuzzy Method of Accuracy Assessment in Experiments 153
4.8 Chapter Conclusions 159
References 159
5 Subpixel Mapping Technique of HSI 161
5.1 Subpixel Mapping for a Land Class with Linear Features Using a Least Square Support Vector Machine (LSSVM) 163
5.1.1 Subpixel Mapping Based on the Least Square Support Vector Machine (LSSVM) 164
5.1.2 Artificially Synthesized Training Samples 166
5.2 Spatial Attraction-Based Subpixel Mapping (SPSAM) 168
5.2.1 Subpixel Mapping Based on the Modified Subpixel/Pixel Spatial Attraction Model (MSPSAM) 168
5.2.2 Subpixel Mapping Based on the Mixed Spatial Attraction Model (MSAM) 172
5.3 Subpixel Mapping Using Markov Random Field with Subpixel Shifted Remote Sensing Images 177
5.3.1 Markov Random Field-Based Subpixel Mapping 177
5.3.2 Markov Random Field-Based Subpixel Mapping with Subpixel Shifted Remote-Sensing Images 181
5.4 Accuracy Assessment 184
5.4.1 Subpixel Mapping for Land Class with Linear Features Using the Least Squares Support Vector Machine (LSSVM) 184
5.4.2 MSPSAM and MSAM 187
5.4.3 MRF-Based Subpixel Mapping with Subpixel Shifted Remote-Sensing Images 192
5.5 Chapter Conclusions 197
References 198
6 Super-Resolution Technique of HSI 200
6.1 POCS Algorithm-Based Super-Resolution Recovery 200
6.1.1 Basic Theory of POCS 200
6.1.2 POCS Algorithm-Based Super-Resolution Recovery 202
6.2 MAP Algorithm-Based Super-Resolution Recovery 206
6.2.1 Basic Theory of MAP 206
6.2.2 MAP Algorithm-Based Super-Resolution Recovery 210
6.3 Resolution Enhancement Method for Single Band 212
6.3.1 Construction of Geometric Dual Model and Interpolation Method 213
6.3.2 Mixed Interpolation Method 216
6.4 Performance Assessment 219
6.4.1 POCS and MAP-Based Super-Resolution Methods 219
6.4.2 Dual Interpolation Method 222
6.5 Chapter Conclusions 228
References 229
7 Anomaly Detection Technique of HSI 230
7.1 Kernel Detection Algorithm Based on the Theory of the Morphology 230
7.1.1 Band Selection Based on Morphology 231
7.1.2 Kernel RX Algorithm Based on Morphology 234
7.2 Adaptive Kernel Anomaly Detection Algorithm 237
7.2.1 The Method of Support Vector Data Description 238
7.2.2 Adaptive Kernel Anomaly Detection Algorithm 241
7.3 Construction of Spectral Similarity Measurement Kernel in Kernel Anomaly Detection 245
7.3.1 The Limitations of Gaussian Radial Basis Kernel 246
7.3.2 Spectral Similarity Measurement Kernel Function 247
7.4 Performance Assessment 251
7.4.1 Effect Testing of Morphology-Based Kernel Detection Algorithm 251
7.4.2 Effect Testing of Adaptive Kernel Anomaly Detection Algorithm 254
7.4.3 Effect Testing of Spectral Similarity Measurement Kernel-Based Anomaly Detection Algorithm 257
7.5 Introduction of Other Anomaly Detection Algorithms 262
7.5.1 Spatial Filtering-Based Kernel RX Anomaly Detection Algorithm 262
7.5.2 Multiple Window Analysis-Based Kernel Detection Algorithm 265
7.6 Summary 268
References 269
8 Dimensionality Reduction and Compression Technique of HSI 270
8.1 Dimensionality Reduction Technique 270
8.1.1 SVM-Based Band Selection 270
8.1.2 Application of Typical Endmember Methods-based Band Selection 275
8.1.3 Simulation Experiments 277
8.2 Compression Technique 279
8.2.1 Vector Quantization-based Compression Algorithm 279
8.2.2 Lifting Scheme-based Compression Algorithm 286
8.3 Chapter Conclusions 292
References 293
9 Introduction of Hyperspectral Remote Sensing Applications 295
9.1 Agriculture 295
9.1.1 Wheat 295
9.1.2 Paddy 297
9.1.3 Soybean 297
9.1.4 Maize 298
9.2 Forest 298
9.2.1 Forest Investigation 298
9.2.2 Forest Biochemical Composition and Forest Health Status 301
9.2.3 Forest Disaster 302
9.2.4 Exotic Species Monitoring 303
9.3 Meadow 303
9.3.1 Biomass Estimation in Meadow 304
9.3.2 Grassland Species Identification 305
9.3.3 Chemical Constituent Estimation 306
9.4 Ocean 307
9.4.1 Basic Research on Ocean Remote Sensing 307
9.4.2 Application Research on Resource and Environment Monitoring of Ocean and Coastal Zone 308
9.4.3 International Development Trend 309
9.5 Geology 310
9.5.1 Mineral Identification 311
9.5.2 Resource Exploration 312
9.6 Environment 316
9.6.1 Atmospheric Pollution Monitoring 316
9.6.2 Soil Erosion Monitoring 317
9.6.3 Water Environment Monitoring 317
9.7 Military Affairs 318
References 320
Appendix 321