Details
Change Detection and Image Time Series Analysis 2
Supervised Methods1. Aufl.
139,99 € |
|
Verlag: | Wiley |
Format: | EPUB |
Veröffentl.: | 01.12.2021 |
ISBN/EAN: | 9781119882282 |
Sprache: | englisch |
Anzahl Seiten: | 272 |
DRM-geschütztes eBook, Sie benötigen z.B. Adobe Digital Editions und eine Adobe ID zum Lesen.
Beschreibungen
<i>Change Detection and Image Time Series Analysis 2</i> presents supervised machine-learning-based methods for temporal evolution analysis by using image time series associated with Earth observation data. Chapter 1 addresses the fusion of multisensor, multiresolution and multitemporal data. It proposes two supervised solutions that are based on a Markov random field: the first relies on a quad-tree and the second is specifically designed to deal with multimission, multifrequency and multiresolution time series.<br /><br />Chapter 2 provides an overview of pixel based methods for time series classification, from the earliest shallow learning methods to the most recent deep-learning-based approaches.<br /><br />Chapter 3 focuses on very high spatial resolution data time series and on the use of semantic information for modeling spatio-temporal evolution patterns.<br /><br />Chapter 4 centers on the challenges of dense time series analysis, including pre processing aspects and a taxonomy of existing methodologies. Finally, since the evaluation of a learning system can be subject to multiple considerations,<br /><br />Chapters 5 and 6 offer extensive evaluations of the methodologies and learning frameworks used to produce change maps, in the context of multiclass and/or multilabel change classification issues.
<p>Contents</p> <p>Preface ix</p> <p>Abdourrahmane M. ATTO, Francesca BOVOLO and Lorenzo BRUZZONE</p> <p>List of Notations</p> <p><b>Chapter 1 Hierarchical Markov Random Fields for High Resolution Land Cover Classification of Multisensor and Multiresolution Image Time Series 1</b></p> <p>Ihsen HEDHLI, Gabriele MOSER, Sebastiano B. SERPICO</p> <p>and Josiane ZERUBIA</p> <p>1.1. Introduction 1</p> <p>1.1.1. The role of multisensor data in time series classification 1</p> <p>1.1.2. Multisensor and multiresolution classification 2</p> <p>1.1.3.Previouswork 5</p> <p>1.2. Methodology 9</p> <p>1.2.1. Overview of the proposed approaches 9</p> <p>1.2.2. Hierarchical model associated with the first proposed method 10</p> <p>1.2.3. Hierarchical model associated with the second proposed method 13</p> <p>1.2.4. Multisensor hierarchical MPM inference 14</p> <p>1.2.5. Probability density estimation through finite mixtures 17</p> <p>1.3.Examplesofexperimentalresults 19</p> <p>1.3.1.Resultsofthefirstmethod 19</p> <p>1.3.2.Resultsofthesecondmethod 22</p> <p>1.4.Conclusion 26</p> <p>xiii</p> <p>1.5.Acknowledgments 26</p> <p>1.6.References 27</p> <p><b>Chapter 2 Pixel-based Classification Techniques for Satellite Image Time Series 33</b></p> <p>Charlotte PELLETIER and Silvia VALERO</p> <p>2.1. Introduction 33</p> <p>2.2. Basic concepts in supervised remote sensing classification 35</p> <p>2.2.1. Preparing data before it is fed into classification algorithms 35</p> <p>2.2.2. Key considerations when training supervised classifiers 39</p> <p>2.2.3. Performance evaluation of supervised classifiers 41</p> <p>2.3.Traditionalclassificationalgorithms 45</p> <p>2.3.1. Support vector machines 45</p> <p>2.3.2. Random forests 51</p> <p>2.3.3. k-nearest neighbor 56</p> <p>2.4. Classification strategies based on temporal feature representations 59</p> <p>2.4.1. Phenology-based classification approaches 60</p> <p>2.4.2 Dictionary-based classificationapproaches 61</p> <p>2.4.3 Shapelet-based classificationapproaches 62</p> <p>2.5.Deeplearningapproaches 63</p> <p>2.5.1. Introduction to deep learning 64</p> <p>2.5.2.Convolutionalneuralnetworks 68</p> <p>2.5.3.Recurrentneuralnetworks 71</p> <p>2.6.References 75</p> <p><b>Chapter 3 Semantic Analysis of Satellite Image Time Series 85</b></p> <p>Corneliu Octavian DUMITRU and Mihai DATCU</p> <p>3.1. Introduction 85</p> <p>3.1.1.TypicalSITSexamples 89</p> <p>3.1.2. Irregular acquisitions 90</p> <p>3.1.3.Thechapterstructure 96</p> <p>3.2.WhyaresemanticsneededinSITS? 96</p> <p>3.3.Similaritymetrics 97</p> <p>3.4. Feature methods 98</p> <p>3.5. Classification methods 98</p> <p>3.5.1.Activelearning 99</p> <p>3.5.2.Relevancefeedback 100</p> <p>3.5.3. Compression-based pattern recognition 100</p> <p>3.5.4.LatentDirichletallocation 101</p> <p>3.6.Conclusion 102</p> <p>vii</p> <p>3.7.Acknowledgments 105</p> <p>3.8.References 105</p> <p><b>Chapter 4 Optical Satellite Image Time Series Analysis for Environment Applications: From Classical Methods to Deep Learning and Beyond 109</b></p> <p>Matthieu MOLINIER, Jukka MIETTINEN,DinoIENCO,ShiQIU and Zhe ZHU</p> <p>4.1. Introduction 109</p> <p>4.2. Annual time series 111</p> <p>4.2.1. Overview of annual time series methods 111</p> <p>4.2.2 Examples of annual times series analysis applications for environmentalmonitoring 112</p> <p>4.2.3.Towardsdensetimeseriesanalysis 116</p> <p>4.3. Dense time series analysis using all available data 117</p> <p>4.3.1. Making dense time series consistent 118</p> <p>4.3.2. Change detection methods 121</p> <p>4.3.3.Summaryandfuturedevelopments 125</p> <p>4.4. Deep learning-based time series analysis approaches 126</p> <p>4.4.1 Recurrent Neural Network (RNN) for Satellite Image TimeSeries 129</p> <p>4.4.2 Convolutional Neural Networks (CNN) for Satellite Image TimeSeries 131</p> <p>4.4.3. Hybrid models: Convolutional Recurrent Neural Network (ConvRNN) models for Satellite Image Time Series 134</p> <p>4.4.4. Synthesis and future developments 136</p> <p>4.5. Beyond satellite image time series and deep learning: convergence between time series and video approaches 136</p> <p>4.5.1 Increased image acquisition frequency: from time series to spacebornetime-lapseandvideos 137</p> <p>4.5.2. Deep learning and computer vision as technology enablers 138</p> <p>4.5.3.Futuresteps 139</p> <p>4.6.References 140</p> <p><b>Chapter 5 A Review on Multi-temporal Earthquake Damage Assessment Using Satellite Images 155</b></p> <p>Gülşen TAŞKIN, EsraERTEN and Enes Oğuzhan ALATAŞ</p> <p>5.1. Introduction 155</p> <p>5.1.1. Research methodology and statistics 159</p> <p>5.2. Satellite-based earthquake damage assessment 165</p> <p>5.3. Pre-processing of satellite images before damage assessment 167</p> <p>5.4. Multi-source image analysis 168</p> <p>5.5. Contextual feature mining for damage assessment 169</p> <p>5.5.1.Texturalfeatures 170</p> <p>5.5.2. Filter-based methods 173</p> <p>5.6. Multi-temporal image analysis for damage assessment 175</p> <p>5.6.1. Use of machine learning in damage assessment problem 176</p> <p>5.6.2. Rapid earthquake damage assessment 180</p> <p>5.7. Understanding damage following an earthquake using satellite-based SAR 181</p> <p>5.7.1. SAR fundamental parameters and acquisition vector 185</p> <p>5.7.2. Coherent methods for damage assessment 188</p> <p>5.7.3. Incoherent methods for damage assessment 192</p> <p>5.7.4. Post-earthquake-only SAR data-based damage assessment 195</p> <p>5.7.5 Combination of coherent and incoherent methods for damage assessment 196</p> <p>5.7.6.Summary 198</p> <p>5.8. Use of auxiliary data sources 200</p> <p>5.9.Damagegrades 200</p> <p>5.10.Conclusionanddiscussion 203</p> <p>5.11.References 205</p> <p><b>Chapter 6 Multiclass Multilabel Change of State Transfer Learning from Image Time Series 223</b></p> <p>Abdourrahmane M. ATTO,HélaHADHRI, FlavienVERNIER</p> <p>and Emmanuel TROUVÉ</p> <p>6.1. Introduction 223</p> <p>6.2. Coarse- to fine-grained change of state dataset 225</p> <p>6.3. Deep transfer learning models for change of state classification 232</p> <p>6.3.1.Deeplearningmodellibrary 232</p> <p>6.3.2.GraphstructuresfortheCNNlibrary 234</p> <p>6.3.3. Dimensionalities of the learnables for the CNN library 236</p> <p>6.4.Changeofstateanalysis 237</p> <p>6.4.1 Transfer learning adaptations for the change of state classificationissues 238</p> <p>6.4.2.Experimentalresults 239</p> <p>6.5.Conclusion 243</p> <p>6.6.Acknowledgments 244</p> <p>6.7.References 244</p> <p>List of Authors 247</p> <p>Index 249</p> <p>Summary of Volume 1 253</p>
<b>Abdourrahmane M. Atto</b> is Associate Professor at the University Savoie Mont Blanc, France. His research interests include mathematical methods and models for artificial intelligence and image time series.<br /><br /><b>Francesca Bovolo</b> is the Head of the Remote Sensing for Digital Earth Unit, Fondazione Bruno Kessler, Italy. Her research interests include remote sensing image time series analysis, content-based time series retrieval and radar sounders.<br /><br /><b>Lorenzo Bruzzone</b> is Professor of Telecommunications and the Founder and Director of the Remote Sensing Laboratory at the University of Trento, Italy. His research interests include remote sensing, machine learning and pattern recognition.