Optical Remote Sensing (eBook)

Advances in Signal Processing and Exploitation Techniques

Saurabh Prasad, Lori M. Bruce, Jocelyn Chanussot (Herausgeber)

eBook Download: PDF

2011 | 2011
VIII, 344 Seiten
Springer Berlin (Verlag)
978-3-642-14212-3 (ISBN)

Infos

Optical remote sensing relies on exploiting multispectral and hyper spectral imagery possessing high spatial and spectral resolutions respectively. These modalities, although useful for most remote sensing tasks, often present challenges that must be addressed for their effective exploitation. This book presents current state-of-the-art algorithms that address the following key challenges encountered in representation and analysis of such optical remotely sensed data. Challenges in pre-processing images, storing and representing high dimensional data, fusing different sensor modalities, pattern classification and target recognition, visualization of high dimensional imagery.

Inhalt

Optical Remote Sensing 2
Preface 4
Contents 6
1 Introduction 8
1…Optical Remote Sensing: The Processing Chain 9
2…Optical Remote Sensing: Key Challenges for Signal Processing and Effective Exploitation 11
References 14
2 Hyperspectral Data Compression Tradeoff 16
Abstract 16
1…Introduction 16
2…Data Acquisition Process and Compression Properties 17
2.1 Data Acquisition Process 17
2.2 Lossy, Lossless, Near-Lossless 19
2.3 Onboard 20
2.4 Image Distribution 20
2.5 Data Availability 22
3…Trends in Compression Algorithms 22
3.1 Prediction-Based 23
3.2 Vector Quantization 24
3.3 Transform Methods 25
3.3.1 Transform 25
3.3.2 Coding 26
3.4 Lossy to Lossless 28
3.5 What is in Use Now? 29
4…Ensuring Sufficient Quality 29
4.1 Why Bothering with Lossy Compression? 29
4.2 Quality Evaluation 30
4.3 Making Comparison Easier 32
5…Reference Results 32
6…Conclusion 34
Acknowledgments 34
References 34
3 Reconstructions from Compressive Random Projections of Hyperspectral Imagery 37
Abstract 37
1…Introduction 38
2…Compressive-Projection Principal Component Analysis (CPPCA) 39
2.1 Overview of CPPCA 40
2.2 The CPPCA Algorithm 41
2.2.1 Eigenvector Recovery 42
2.2.2 Coefficient Recovery 43
3…Compressed Sensing (CS) 44
4…Empirical Comparisons on Hyperspectral Imagery 45
4.1 Performance of Single-Task and Multi-Task CS 46
4.2 Performance of CPPCA and CS 48
4.3 Execution Times 51
5…Conclusions 52
References 52
4 Integrated Sensing and Processing for Hyperspectral Imagery 55
Abstract 55
1…Introduction 56
2…Variable Resolution Hyperspectral Sensing 58
2.1 Mathematical Representation 58
2.2 Reduced Resolution Imaging 61
3…Experimental Results 65
3.1 Improving SNR Using Hadamard Multiplexing 65
3.2 Variable Resolution Hyperspectral Sensing 67
4…Summary 69
References 70
5 Color Science and Engineering for the Display of Remote Sensing Images 71
Abstract 71
1…Introduction 71
2…Challenges 72
2.1 Information Loss 72
2.2 Metrics 73
2.3 Visual Interpretation 73
2.4 Color Saturation and Neutrals 75
2.5 Color Blindness 76
2.6 Case Study: Principal Components Analysis for Visualization 76
3…Some Solutions 77
3.1 Optimized Basis Functions 78
3.2 Adapting Basis Functions 80
3.3 White Balance 83
4…Conclusions and Open Questions 84
References 85
6 An Evaluation of Visualization Techniques for Remotely Sensed Hyperspectral Imagery 86
Abstract 86
1…Introduction 86
2…Image Construction 88
3…Comparative Visualization Techniques 89
3.1 Hard Classification Visualization 90
3.2 Soft Classification Visualization 90
3.3 Double Layer Visualization 91
4…Experimental Design and Settings 92
5…Experimental Tasks and Results 94
5.1 Global Pattern Display Capability 94
5.1.1 Perceptual Edge Detection 94
Task 95
Results 95
5.1.2 Block Value Estimation 96
Task 96
Result 97
5.2 Ability to Convey Local Information 97
5.2.1 Class Recognition 98
Task 98
Results 98
5.2.2 Target Value Estimation 100
Task 100
Results 100
6…Discussion and Conclusions 101
Acknowledgments 102
References 102
7 A Divide-and-Conquer Paradigm for Hyperspectral Classification and Target Recognition 104
Abstract 104
1…Introduction 105
2…The Proposed Framework 107
2.1 Subspace Identification: Partitioning the Hyperspectral Space 108
2.2 Pre-processing at the Subspace Level 111
2.2.1 Linear Discriminant Analysis (LDA) 111
2.2.2 Kernel Discriminant Analysis (KDA) 112
2.3 Classifier 114
2.4 Decision Fusion 115
3…Experimental Hyperspectral Datasets 116
3.1 Handheld Hyperspectral Data 116
3.2 Airborne Hyperspectral Data 117
4…Experimental Setup and Results 118
4.1 Experiments with Handheld HSI Data 119
4.1.1 Experiment 1: MCDF with LDA Based Pre-processing at the Subspace Level 119
4.1.2 Experiment 2: MCDF with KDA Based Pre-processing at the Subspace Level 120
4.2 Experiments with Aerial HSI Data 123
5…Conclusions, Caveats and Future Work 124
Acknowledgments 125
References 125
8 The Evolution of the Morphological Profile: from Panchromatic to Hyperspectral Images 128
Abstract 128
1…Introduction 129
2…Preliminaries of Mathematical Morphology 130
2.1 Fundamental Properties 130
2.2 Opening and Closing by Reconstruction 131
2.3 Attribute Filters 133
3…Morphological Profiles for the Analysis of Panchromatic Images 136
3.1 Morphological Profiles 136
3.2 Attribute Profiles 138
3.3 Experimental Results and Discussion 139
4…Extended Morphological Profiles to the Analysis of Multispectral and Hyperspectral Images 142
4.1 Problem of Extending the Morphological Operators to Multi-tone Images 142
4.2 Extended Morphological Profile 143
4.3 Extended Attribute Profiles 144
4.4 Experimental Results and Discussion 145
5…Conclusion 148
Acknowledgments 149
References 149
9 Decision Fusion of Multiple Classifiers for Vegetation Mapping and Monitoring Applications by Means of Hyperspectral Data 152
Abstract 152
1…Introduction 153
2…Study Area 154
2.1 Image Data 154
2.2 Field Data 155
2.3 Training Data 155
3…Methods for Fusion of Multiple Classifiers 157
3.1 Decision Fusion Using Hierarchical Tree Structure 157
3.2 Decision Fusion Using the Hierarchical Tree and Class Membership Values 159
3.3 Class-Dependent Neural Networks Ensemble 160
4…Accuracy Assessment 161
4.1 Comparison of Classification Results 162
5…Results 163
5.1 Results of Various Tested Classifiers 163
5.2 Results of Class Dependent Neural Networks 168
5.3 Results of Decision Fusion Using Hierarchical Tree Structure 169
5.4 Results of Hierarchical Tree Coupled with Probability Labels 171
5.5 The Assessment of Significance of the Accuracy Values 172
6…Conclusions 173
Acknowledgments 174
References 174
10 A Review of Kernel Methods in Remote Sensing Data Analysis 176
Abstract 176
1…Introduction 177
1.1 Classification with Kernels 177
1.2 Model Inversion with Kernels 178
1.3 Feature Extraction with Kernels 178
2…Introduction to Kernel Methods 179
2.1 Measuring Similarity with Kernels 179
2.2 Positive Definite Kernels 179
2.3 Basic Operations with Kernels 180
2.4 Standard Kernels 180
2.5 Kernel Development 181
3…Kernel Methods in Remote Sensing Data Classification 182
3.1 Support Vector Machine 182
3.2 nu -Support Vector Machine 183
3.3 Support Vector Data Description 185
3.4 One-Class Support Vector Machine 185
3.5 Kernel Fisher’s Discriminant 186
3.6 Experimental Results for Supervised Classification 187
3.6.1 Linear versus nonlinear 188
3.6.2 {/varvec /nu} -SVM versus OC-SVM 188
3.6.3 Support Vector versus Fisher’s Discriminant 189
3.7 Semisupervised Image Classification 189
3.7.1 Manifold-Based Regularization Framework 190
3.7.2 Semisupervised Regularization Framework 190
3.7.3 Laplacian Support Vector Machine 191
3.7.4 Transductive SVM 192
3.8 Experimental Results for Semisupervised Classification 192
4…Kernel Methods in Biophysical Parameter Estimation 193
4.1 Support Vector Regression 194
4.2 Relevance Vector Machines 195
4.3 Gaussian Processes 197
4.4 Experimental Results 198
5…Kernel Methods for Feature Extraction 199
5.1 Mutivariate Analysis Methods 200
5.1.1 Principal Component Analysis 200
5.1.2 Partial Least Squares 201
5.2 Kernel Multivariate Analysis 201
5.2.1 Kernel Principal Component Analysis 202
5.2.2 Kernel Partial Least Squares 203
5.3 Experimental Results 203
6…Future Trends in Remote Sensing Kernel Learning 204
6.1 Multiple Kernel Learning 204
6.2 Transfer Learning 204
6.3 Structured Learning 205
6.4 Active Learning 205
6.5 Parallel Implementations 205
7…Conclusions 206
Acknowledgments 206
References 206
11 Exploring Nonlinear Manifold Learning for Classification of Hyperspectral Data 212
Abstract 212
1…Introduction 213
2…Nonlinear Manifold Learning for Dimensionality Reduction 214
2.1 Dimensionality Reduction Within a Graph Embedding Framework 215
2.2 Global Manifold Learning 215
2.2.1 Isometric Feature Mapping (Isomap) 216
2.2.2 Kernel Principal Component Analysis (KPCA) 216
2.3 Local Manifold Learning 217
2.3.1 Locally Linear Embedding (LLE) 217
2.3.2 Local Tangent Space Alignment (LTSA) 218
2.3.3 Laplacian Eigenmaps (LE) 218
2.4 Supervised Local Manifold Learning 219
2.5 Kernel-Based Out-of-Sample Extension 219
3…Remotely Sensed Data for Comparative Experiments 220
3.1 Botswana Hyperion Data (BOT) 220
3.2 Kennedy Space Center AVIRIS Data (KSC) 221
3.3 Indian Pine AVIRIS Data (IND PINE) 222
3.4 ACRE ProspectTIR Data (ACRE) 222
4…Experimental Results 222
4.1 Performance of Dimensionality Reduction Methods (DR) for BOT Hyperion Data 223
4.2 Comparison of DR Methods for BOT, KSC, IND PINE, and ACRE Sites 227
4.3 Manifold Coordinates for DR Methods 231
5…Summary and Conclusions 234
Acknowledgment 237
References 237
12 Recent Developments in Endmember Extraction and Spectral Unmixing 240
Abstract 240
1…Introduction 241
2…Linear Spectral Unmixing 243
2.1 Problem Formulation 243
2.2 Endmember Extraction 244
2.2.1 N-FINDR 245
2.2.2 Orthogonal Subspace Projection (OSP) 246
2.2.3 Vertex Component Analysis (VCA) 247
2.2.4 Automatic Morphological Endmember Extraction (AMEE) 247
2.2.5 Spatial Spectral Endmember Extraction (SSEE) 248
2.2.6 Spatial Pre-Processing (SPP) 250
2.3 Unconstrained Versus Constrained Linear Spectral Unmixing 251
3…Nonlinear Spectral Unmixing 252
3.1 Problem Formulation 252
3.2 Neural Network-Based Spectral Unmixing 253
3.3 Automatic Selection and Labeling of Training Samples 255
4…Experimental Results 256
4.1 First Experiment: AVIRIS Hyperspectral Data 256
4.2 Second Experiment: DAIS 7915 and ROSIS Hyperspectral Data 259
4.2.1 Data Description 260
4.2.2 Fractional Abundance Estimation Results 262
5…Parallel Implementation Case Study 264
6…Conclusions and Future Research 268
Acknowledgements 269
References 269
13 Change Detection in VHR Multispectral Images: Estimation and Reduction of Registration Noise Effects 273
Abstract 273
1…Introduction 274
2…Notation and Background 275
3…Analysis of Registration Noise Properties 277
3.1 Experimental Setup 278
3.1.1 Experiment 1: Effects of Increasing Misregistration on Unchanged Pixels 279
3.1.2 Experiment 2: Effects of Increasing Misregistration on Changed Pixels 280
3.1.3 Experiment 3: Effects of Misregistration at Different Scales 280
3.2 Properties of RN in VHR Images 282
4…Proposed Technique for the Adaptive Estimation of the Registration Noise Distribution 287
5…Proposed Change-Detection Technique Robust to Registration Noise 290
5.1 Registration Noise Identification 291
5.2 Context-Sensitive Decision Strategy for the Generation of the Final Change-Detection Map 293
6…Experimental Results 294
6.1 Data Set Description 294
6.2 Estimation Results 296
6.3 Change-Detection Results 298
7…Discussion and Conclusion 301
References 302
14 Effects of the Spatial Enhancement of Hyperspectral Images on the Distribution of Spectral Classes 304
Abstract 304
1…Introduction 304
2…Spatial Enhancement of Hyperspectral Images 306
2.1 Component Substitution Methods 306
2.2 Multiresolution Methods 307
2.3 Selected Methods for Testing on HS+Pan Images 308
2.4 Evaluation of Spatial Enhancement Methods 310
3…Dimensionality Reduction for the Assessment of Pan-Sharpening Algorithms 310
4…Experimental Results 313
4.1 ‘Visual’ Approach 313
4.1.1 GIHS 313
4.1.2 HPF 314
4.1.3 HPF-P 314
4.1.4 GMMSE 316
4.2 Linear and Non-Linear Sample Similarity Measures 316
4.3 PCA 318
4.4 Kernel PCA 321
4.5 Linearity Preserving Projection (LPP) 324
5…Conclusions 328
References 328
15 Fusion of Optical and SAR Data for Seismic Vulnerability Mapping of Buildings 331
Abstract 331
1…Introduction 332
2…Fusion of Optical and Radar Data 332
3…Data Fusion for Vulnerability Assessment 334
3.1 The Aim 334
3.2 Remote Sensing as a Tool 335
3.3 Decision-Level Fusion 340
4…Conclusions 341
Acknowledgments 342
References 342

Produkt-Details

• Erscheint lt. Verlag: 23.3.2011
• Reihe/Serie: Augmented Vision and Reality
• Zusatzinfo: VIII, 344 p.
• Verlagsort: Berlin
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik
• Themenwelt: Technik ► Elektrotechnik / Energietechnik
• Themenwelt: Technik ► Nachrichtentechnik
• Schlagworte: high spectral resolution • hyperspectral imagery possessing • multispectral imagery possessing • pre-processing images
• ISBN-10: 3-642-14212-5 / 3642142125
• ISBN-13: 978-3-642-14212-3 / 9783642142123