Оглавление
Группа авторов. Change Detection and Image Time Series Analysis 2
Table of Contents
Guide
Pages
Change Detection and Image Time Series Analysis 2. Supervised Methods
Preface
Volume 1: Unsupervised methods
Volume 2: Supervised methods
List of Notations
1. Hierarchical Markov Random Fields for High Resolution Land Cover Classification of Multisensor and Multiresolution Image Time Series
1.1. Introduction. 1.1.1. The role of multisensor data in time series classification
1.1.2. Multisensor and multiresolution classification
1.1.3. Previous work
1.2. Methodology. 1.2.1. Overview of the proposed approaches
1.2.2. Hierarchical model associated with the first proposed method
1.2.3. Hierarchical model associated with the second proposed method
1.2.4. Multisensor hierarchical MPM inference
1.2.4.1. Initializing on the first quad-tree
1.2.4.2. First top-down pass
1.2.4.3. Bottom-up pass
1.2.4.4. Second top-down pass
1.2.4.5. Generation of the output map
1.2.5. Probability density estimation through finite mixtures
1.3. Examples of experimental results. 1.3.1. Results of the first method
1.3.2. Results of the second method
1.4. Conclusion
1.5. Acknowledgments
1.6. References
2. Pixel-based Classification Techniques for Satellite Image Time Series
2.1. Introduction
2.2. Basic concepts in supervised remote sensing classification
2.2.1. Preparing data before it is fed into classification algorithms
2.2.1.1. Satellite data
2.2.1.2. Reference labels
2.2.1.3. Split procedure into training, testing and validation datasets
2.2.2. Key considerations when training supervised classifiers
2.2.2.1. Quality of training data
2.2.2.2. The curse of dimensionality
2.2.2.3. How much training data do I need?
2.2.2.4. Overfitting, underfitting and the bias-variance trade-off
2.2.3. Performance evaluation of supervised classifiers
2.2.3.1. Classical evaluation metrics
2.2.3.2. Evaluation of uncertainty in classification performances
2.3. Traditional classification algorithms
2.3.1. Support vector machines
2.3.1.1. Hard-margin linear SVM for linearly separable data
2.3.1.2. Soft-margin linear SVM for almost linearly separable data
2.3.1.3. Nonlinear SVM and the kernel trick
2.3.1.4. From binary to multiclass problems
2.3.1.5. Hyperparameter tuning
2.3.2. Random forests
2.3.2.1. Binary decision tree
2.3.2.2. From tree to forest
2.3.2.3. Hyperparameter setting
2.3.2.4. Random forest for data analysis
2.3.3. k-nearest neighbor
2.3.3.1. The k-nearest neighbor algorithm
2.3.3.2. Dynamic time warping
2.4. Classification strategies based on temporal feature representations
2.4.1. Phenology-based classification approaches
2.4.2. Dictionary-based classification approaches
2.4.3. Shapelet-based classification approaches
2.5. Deep learning approaches
2.5.1. Introduction to deep learning
2.5.1.1. Multilayer perceptron
2.5.1.2. Basic knowledge for training a deep neural network
2.5.2. Convolutional neural networks
2.5.3. Recurrent neural networks
2.5.3.1. A simple recurrent neural network model
2.5.3.2. RNN variants
2.6. References
3. Semantic Analysis of Satellite Image Time Series
3.1. Introduction
3.1.1. Typical SITS examples
3.1.2. Irregular acquisitions
3.1.3. The chapter structure
3.2. Why are semantics needed in SITS?
3.3. Similarity metrics
3.4. Feature methods
3.5. Classification methods
3.5.1. Active learning
3.5.2. Relevance feedback
3.5.3. Compression-based pattern recognition
3.5.4. Latent Dirichlet allocation
3.6. Conclusion
3.7. Acknowledgments
3.8. References
4. Optical Satellite Image Time Series Analysis for Environment Applications: From Classical Methods to Deep Learning and Beyond
4.1. Introduction
4.2. Annual time series. 4.2.1. Overview of annual time series methods
4.2.2. Examples of annual times series analysis applications for environmental monitoring
4.2.3. Towards dense time series analysis
4.3. Dense time series analysis using all available data
4.3.1. Making dense time series consistent
4.3.1.1. Atmospheric correction
4.3.1.2. Automated cloud and cloud shadow detection
4.3.1.3. Harmonization of multiple sensors
4.3.2. Change detection methods
4.3.2.1. Map classification
4.3.2.2. Trajectory classification
4.3.2.3. Statistical boundary
4.3.2.4. Ensemble approach
4.3.2.5. Two of the widely used algorithms
4.3.3. Summary and future developments
4.4. Deep learning-based time series analysis approaches
4.4.1. Recurrent Neural Network (RNN) for Satellite Image Time Series
4.4.2. Convolutional Neural Networks (CNN) for Satellite Image Time Series
4.4.3. Hybrid models: Convolutional Recurrent Neural Network (ConvRNN) models for Satellite Image Time Series
4.4.4. Synthesis and future developments
4.5. Beyond satellite image time series and deep learning: convergence between time series and video approaches
4.5.1. Increased image acquisition frequency: from time series to spaceborne time-lapse and videos
4.5.2. Deep learning and computer vision as technology enablers
4.5.3. Future steps
4.6. References
5. A Review on Multi-temporal Earthquake Damage Assessment Using Satellite Images
5.1. Introduction
5.1.1. Research methodology and statistics
5.2. Satellite-based earthquake damage assessment
5.3. Pre-processing of satellite images before damage assessment
5.4. Multi-source image analysis
5.5. Contextual feature mining for damage assessment
5.5.1. Textural features
5.5.2. Filter-based methods
5.6. Multi-temporal image analysis for damage assessment
5.6.1. Use of machine learning in damage assessment problem
5.6.2. Rapid earthquake damage assessment
5.7. Understanding damage following an earthquake using satellite-based SAR
5.7.1. SAR fundamental parameters and acquisition vector
5.7.2. Coherent methods for damage assessment
5.7.3. Incoherent methods for damage assessment
5.7.4. Post-earthquake-only SAR data-based damage assessment
5.7.5. Combination of coherent and incoherent methods for damage assessment
5.7.6. Summary
5.8. Use of auxiliary data sources
5.9. Damage grades
5.10. Conclusion and discussion
5.11. References
6. Multiclass Multilabel Change of State Transfer Learning from Image Time Series
6.1. Introduction
6.2. Coarse- to fine-grained change of state dataset
6.3. Deep transfer learning models for change of state classification. 6.3.1. Deep learning model library
6.3.2. Graph structures for the CNN library
6.3.3. Dimensionalities of the learnables for the CNN library
6.4. Change of state analysis
6.4.1. Transfer learning adaptations for the change of state classification issues
6.4.2. Experimental results
6.5. Conclusion
6.6. Acknowledgments
6.7. References
List of Authors
Index. A
B
C
D
E
F
G
H, I
J, K, L
M
N
O, P
Q, R
S
T, U
V, W, X
Summary of Volume 1
WILEY END USER LICENSE AGREEMENT
